Research Appraisals
Evidence-based critical appraisals of the latest medical research, systematically evaluated using Oxford CEBM methodology.
Showing 90 appraisals
Journal of chemical information and modeling
Reinforcement Learning-Driven Multiproperty Optimization in Molecular Design Using Multicontext Transcriptome Data
Drug discovery inherently involves multiparameter optimization in the molecular design because drug candidate molecules must meet diverse properties such as bioactivity, synthesizability, and pharmacokinetic properties. This optimization has traditionally relied on iterative manual design and experimental testing, which are labor-intensive and time-consuming. There is therefore a strong incentive to develop computational methods that efficiently design drug-like molecules with multiple favorable properties using chemical and biological data on therapeutic targets. This study proposes a novel computational method for multiproperty optimization in the molecular structure design of bioactive molecules using multicontext (i.e., chemically and genetically perturbed) transcriptome data on human cells. We integrate a molecular generative model conditioned on a transcriptome profile observed with the target gene knockdown or overexpression into a reinforcement learning framework, enabling simultaneous optimization of a quantitative estimate of drug-likeness, synthetic accessibility score, and water/octanol partition coefficient, while accounting for system-level biological effects on a therapeutic target. Using comprehensive benchmarking against established baselines and rigorous validation across multiple metrics, we demonstrate that the proposed method consistently yields molecules with more favorable drug-like characteristics than existing methods. This proposed method can help achieve more efficient identification of novel drug candidates.
29 July 2026
Read appraisal →Journal of chemical information and modeling
PegaPlus─Interactive Machine Learning by Human Observation for Efficient Clustering and Analysis of Structure-Activity Data
In the early stages of drug discovery, the identification of molecular series with lead-like properties is essential for structure-activity relationship studies. A common first step is to cluster data derived from high-throughput screening (HTS). This data then needs to be exhaustively refined by an expert to identify bioactive compound classes. Here, we present the novel, interactive, and visual approach named PegaPlus for the analysis of HTS data, facilitating the incorporation of expert knowledge with the strategy of learning by observation. PegaPlus uses a tailored variant of a stochastic proximity embedding algorithm to visualize the clustered data in a two-dimensional plane via a web interface. The user can iteratively modify the proposed compound clustering by dragging data points in the drawing plane. At each step, an online support vector machine learns and improves its clustering predictions by incorporating user modifications and the molecular features of the compounds. The visualization is also updated based on the user's input and molecular similarity. PegaPlus was developed with the exploratory process in mind and allows the user to freely organize molecules on the drawing plane, or filter them by various molecular properties. Our evaluation shows that PegaPlus reduces the number of refinement steps needed by half in comparison to the purely manual approach on multiple data sets. These results demonstrate how an interactive machine learning approach can reliably support medicinal chemists and automate time-consuming, manual refinement tasks. The PegaPlus web server is openly available at pegaplus.zbh.uni-hamburg.de.
14 July 2026
Read appraisal →Nucleic acids research
FABIAN-variant 2026: improved prediction of the effects of DNA variants on transcription factor binding
Variants in promoters and enhancers can alter the binding of transcription factors (TFs), but their functional assessment remains difficult. FABIAN-variant is a web application that predicts the effects of DNA variants on TF binding by comparing position weight matrix (PWM) and transcription factor flexible model (TFFM) scores between reference and variant alleles. Here, we present FABIAN-variant 2026, a major update that expands the prediction model library from ~5000 to over 40 000 models for >1500 human TFs, sourced from nine PWM databases and including 1290 TFFMs. The application now supports the mouse genome (GRCm38 and GRCm39) with over 35 000 models for >1100 mouse TFs. An optional BPNet deep learning scorer provides neural network-based binding predictions for 240 human TFs. Known TF binding site information has been expanded from three to five sources. Predictions for over 1400 heterodimer TF complexes have been added. The web server has been rewritten in Rust and the scoring engine optimized, reducing runtime by ~70%. A RESTful JSON API and a standalone command-line version enable programmatic access and local high-throughput analysis. FABIAN-variant 2026 is available at https://fabianapp.org/variant26/. The web server is free and open to all users and there is no login requirement.
13 July 2026
Read appraisal →Briefings in bioinformatics
Literature-informed gene extraction and ranking for multimodal data fusion.
Published biomedical experiments provide an increasingly rich collection of results and identify genes potentially involved in diverse biological mechanisms. However, individual studies are often confined to narrow experimental contexts and are restricted to single omics layers. Cross-study knowledge aggregation can broaden this perspective and enable the construction of global, context-aware gene rankings. Recent developments in natural language processing have made large-scale literature mining increasingly feasible. This enables the systematic extraction and fusion of symbolic knowledge from published experiments. We present pathXcite, a software that extracts genes associated with specific contexts, such as diseases or biological mechanisms from the literature, and ranks them by contextual relevance. These relevance-based gene rankings can compress a scientific context into a symbolic representation. This representation enables diverse downstream analyses, including cross-context comparisons, network-based analysis, enrichment analysis, and integration with experimental omics data. In multiple use cases, we show how our extraction and fusion strategy can be applied to uncover hidden aspects in biological data.
6 July 2026
Read appraisal →Mammalian genome : official journal of the International Mammalian Genome Society
A novel biological function-based method for mining core genes in rare disease with limited cases
Conventional high - throughput sequencing data analysis methods, relying on large cohorts, are unsuitable for rare diseases with few cases and poor at explaining gene biological functions. This study proposes a biological-function-based method for mining core genes under limited-sample conditions. Using colorectal cancer (CRC) and its rare subtype signet-ring cell carcinoma (SRCC) as examples, differential expression analysis was performed separately for each SRCC sample against CRC samples. Shared genes obtained through progressive intersections were defined as the initial core gene set (List 1). Non-core genes were subjected to functional enrichment analysis, and pathway-associated genes were integrated to construct the non-core gene set. Two independent SRCC samples were subsequently used for secondary mining to obtain the final biologically significant core gene set (List 2). In addition, RankProd (RP), Robust Rank Aggregation (RRA), and an external mucinous adenocarcinoma dataset (GSE281917) were used for validation. First-round mining identified 246 core genes in SRCC, while secondary mining generated 66 and 65 biologically significant core genes from two validation samples with a high degree of overlap. RP identified 1,850 candidate genes and RRA identified 601 candidate genes, among which 64 genes overlapped with the 65 genes identified by this method. In the external mucinous adenocarcinoma dataset, first-round mining identified 144 core genes, while secondary mining generated 122 and 119 core genes with high overlap between validation samples. This method enables the identification of biologically significant core genes in rare diseases under limited-sample conditions, improves the biological interpretability of gene-mining results, demonstrates stable performance in both SRCC and external mucinous adenocarcinoma datasets, and provides a reliable experimental basis for the advancement of precision medicine in rare disease research.
15 June 2026
Read appraisal →Science advances
PathTIGR: A pathway topology-informed graph representation learning framework for immunotherapy response prediction
Immunotherapy has revolutionized cancer treatment, yet substantial inter-patient response heterogeneity limits therapeutic benefit to specific patient subsets. Here, we present PathTIGR, a pathway topology-informed graph representation learning framework that systematically integrates biological pathway network topology knowledge with genome variation information for immunotherapy response prediction. PathTIGR uses a three-component design: (i) pathway graph encoder with multihead attention embeding pathway topology knowledge and cancer genomic variants to pathway representation, (ii) transformer module capturing pathway regulatory dependencies, and (iii) multilayer perceptron synthesizing pathway-level representations to predict immunotherapy response. This architecture enables PathTIGR to capture complex molecular interactions underlying immunotherapy response. Comprehensive validation across multiple independent immunotherapy cohorts demonstrates that PathTIGR achieves superior predictive performance compared to established biomarkers and state-of-the-art deep learning approaches while maintaining biological interpretability through identification of key signatures underlying response heterogeneity. PathTIGR represents an interpretable graph-based learning framework that enhances immunotherapy response prediction and elucidates molecular determinants of therapeutic efficacy, thereby facilitating the advancement of precision cancer immunotherapy.
14 June 2026
Read appraisal →Journal of medicinal chemistry
MEG-mod: A Multiview Enhanced Graph Neural Network for Knockdown Efficiency Prediction of Chemically Modified siRNA
Chemical modification is essential for improving the stability and knockdown efficiency of siRNAs. However, the combinatorial complexity of modification types and positions makes rational design difficult. Here, we propose MEG-mod, a deep learning framework for predicting the knockdown efficiency of chemically modified siRNAs. By integrating a literature-reading agent, manual curation, and public database resources, we construct an expanded data set. MEG-mod jointly models sequence context, physicochemical properties, chemical modification features, and duplex structural relationships. A structure-aware Transformer-based graph neural network captures base-pairing and adjacency dependencies within double-stranded siRNAs, while a modification-base fusion module models context-dependent modification effects. MEG-mod outperforms existing methods, achieving a pearson correlation coefficient of 0.9171. The attention mechanism enables interpretable analysis of key modification positions and modification types, and provides recommendations consistent with reported experimental designs. MEG-mod is available as a public web server (https://lmmd.ecust.edu.cn/megmod/), providing a practical tool for prediction and candidate prioritization of chemically modified siRNAs.
13 June 2026
Read appraisal →Medicine
Environmental toxicant glyphosate induces cardiotoxicity: New insights from network toxicology, integrated machine learning, molecular modeling and multidimensional bioinformatics analysis
Glyphosate is one of the world's most widely used herbicide, yet the causal relationship between glyphosate and cardiotoxicity remains inadequate. To elucidate the cardiotoxic mechanisms of glyphosate, an integrative strategy was developed in this work, which combines computational toxicometry, predictive machine learning, and atomic-level molecular modeling. Initially, glyphosate's systematic toxicological profile was evaluated using ProTox-3.0 and ADMETlab 3.0. Candidate genes associated with glyphosate were obtained from 3 separate databases. Target genes associated with cardiotoxicity were integrated from 2 databases. Intersection analysis identified 113 core candidate genes associated with glyphosate and cardiotoxicity. Functional enrichment-related analyses were employed and revealed that the target genes were significantly involved in key biological processes such as ERBB3 signaling pathways. In order to refine the candidate genes to a more precise gene pool, net topology analyses were employed and identified 16 key feature genes. We subsequently integrated machine learning algorithms and validated the results using the GEO database, ultimately identifying 4 hub genes, AKT1, IL1B, BRCA1, and PTGS2. Finally, we applied molecular docking analysis and revealed strong binding affinities, which suggested direct interaction mechanisms in glyphosate-induced cardiotoxicity. In summary, our study establishes a comprehensive mechanistic framework for glyphosate-induced cardiotoxicity.
31 May 2026
Read appraisal →Journal of computer-aided molecular design
Target-aware molecule SMILES generation using a large language model with retrieval-augmented generation, multi-turn memory, and a predictive model
Given the vastness of chemical space and the cost and time requirements of high-throughput screening, resource-efficient computational strategies are needed to prioritize candidate drug molecules. Here, we evaluated whether an open-weight large language model (LLM) augmented with retrieval-augmented generation (RAG), multi-turn memory, and a pretrained drug-target interaction predictor can generate target-aware molecules without task-specific retraining. Protein-ligand-pKi examples retrieved from BindingDB, Davis, and KIBA were used as contextual guidance, while DeepPurpose provided the optimization signal during iterative SMILES refinement. Our approach produced a statistically significant increase in predicted pKi across successive multi-turn memory iterations, indicating that retrieval- and memory-guided refinement may improve target-conditioned molecular generation. The proposed framework also showed favorable molecular generation characteristics, with novelty reaching 100%, diversity up to 0.882, and uniqueness up to 1.0, suggesting that retrieval did not reduce the process to simple reproduction of known ligands but instead supported exploration of new regions of chemical space. In addition, the approach yielded supportive docking results consistent with the generation of chemically relevant candidate ligands. However, the increase in predicted affinity was accompanied by trade-offs in molecular quality, including reduced validity and drug-likeness in later iterations. Taken together, these findings suggest that this framework provides a flexible and comparatively resource-efficient strategy for target-aware de novo molecular design, while further multi-objective optimization and independent validation remain necessary.
31 May 2026
Read appraisal →Cell reports
Artificial intelligence-driven donor-recipient gut microbiome matching for optimized fecal microbiota transplantation
Fecal microbiota transplantation (FMT) has emerged as a promising therapy for gastrointestinal diseases, yet its clinical efficacy remains individually variable. Here, we analyze multi-kingdom and functional profiles in pre- and post-FMT metagenomes from 515 FMTs across 30 cohorts and 12 diseases, in which 94 metagenomes from 44 FMTs are newly collected. We reveal a robust association between clinical efficacy and post-FMT microbiome convergence of recipients toward donors, across diseases. To predict post-FMT microbial convergence, we develop MOZAIC (Microbiome Matching Optimization via Artificial Intelligence), a framework that integrates multi-dimensional donor-recipient microbiota features. MOZAIC achieves an average area under the curve (AUC) of 0.88 and accuracy/recall >0.80 in forecasting microbiome convergence, with 78.7% accuracy in predicting clinical outcomes, and retrospectively simulates a 1.44-fold improvement (from 49.4% to 71.0%) in clinical response rates over baseline. This study establishes microbiome convergence as a key mediator of FMT and provides a scalable tool for precision matching in microbiota-based therapies.
28 May 2026
Read appraisal →Journal of chemical theory and computation
MolXProt: A Cross-Attention Transformer-Based Graph Neural Network for Protein-Ligand Binding Affinity Prediction
Accurate and fast prediction of drug-target binding affinities (DTAs) is key for drug discovery; however, many methods, such as docking and empirical scoring, fail when generalizing to unseen cases. In this study, we introduced the MolXProt architecture, a novel transformer-based graph neural network that integrates graph ligand representations with protein language models using bidirectional multihead cross-attention. Our model is shown to be scalable to over 100000 protein-ligand pairs of mixed data sets, achieving 50% of predictions within ±1.0 kcal/mol and 80% within ±2.0 kcal/mol. We show that the architecture can explicitly learn residue-atom interactions while being computationally friendly via protein-token compression. By mapping token-residue interactions, we demonstrated that the model learns key binding pocket residues in benchmark complexes, such as CDK2-Staurosporine and DHFR-Methotrexate, but under-represents the hydrogen-bonding networks. Our calibration bias analysis revealed that the model overpredicted strong binders and underpredicted weak binders, which are linked to data imbalances and heteroscedastic noise. A simple posthoc isotonic correction partially mitigated the bias. Latent space analysis showed that the model learned continuous binding affinity manifolds without split leakage. Our work highlights a novel architecture that offers unique insights into binding mechanisms via transformer-based cross-attention and is computationally inexpensive.
27 May 2026
Read appraisal →Journal of chemical theory and computation
Predicting and Decoding Allosteric Binding Sites Using Protein Language Models and Structure-Based Machine Learning: An Energy Landscape-Guided Explainable AI Framework
Computational prediction of allosteric binding sites in protein structures remains a persistent challenge, as these regulatory pockets evade detection by both sequence-based and structure-based algorithms. Both computational and physical origins of this predictive asymmetry remain insufficiently understood. In this study, we systematically examine the determinants of binding site predictability using a dual framework that integrates a fine-tuned protein language model and the structure-based method P2Rank as complementary tools probing a diverse data set of 453 human kinases, together with a physics-based interpretability layer derived from energy landscape frustration analysis. Both predictors exhibit a sharp and reproducible dichotomy on protein kinases, in which orthosteric ATP-binding sites can be identified with high precision, whereas allosteric sites are detected with substantially lower confidence across kinase structures and distinct conformational states. To decode this divergence, we deploy energy landscape-based explainable AI approach that integrates local frustration analysis as an independent physical interpretability layer, mapping predictive behavior to the underlying energetic organization of protein structures. This analysis reveals that predictive success is governed by the local energetic embedding of binding sites within the protein energy landscape. Orthosteric pockets are located in minimally frustrated basins that generate strong evolutionary and structural signatures, whereas allosteric pockets occupy predominantly neutrally frustrated zones associated with conformational plasticity and reduced evolutionary constraint. By integrating the prediction results with energy landscape analysis, our framework converts predictive performance into physically interpretable descriptors of binding site organization in protein kinases. These results establish energy landscape frustration as a potentially important determinant of algorithmic visibility and an interpretability layer providing a feasible strategy for diagnosing the limits of current prediction methods.
27 May 2026
Read appraisal →Renal failure
Unveiling the role of air pollution in diabetic kidney disease: an integrated study combining network toxicology, machine learning, and Mendelian randomization
Diabetic kidney disease (DKD) is a major and severe complication associated with diabetes. Air pollution is not only an independent risk factor for metabolic disorders but also an "accelerator" of DKD progression. This study seeks to investigate the molecular pathways connecting air pollution to DKD. Multiple databases were integrated to obtain potential target genes of 10 common air pollutants. The gene expression omnibus (GEO) database was employed to acquire DKD datasets. Differential expression analysis and weighted correlation network analysis (WGCNA) were performed to identify DKD-related genes. 12 machine learning algorithms were utilized to generate 113 unique predictive models, which were employed to select hub genes. Subsequently, MR, single-cell, gene set enrichment analysis (GSEA), and immune infiltration analyses were performed, followed by molecular docking of hub genes with air pollutants. A total of 714 targets were identified from 10 air pollutants, and 80 potential targets were identified from DKD transcriptomic data. Machine learning methods identified 5 hub genes that are closely associated with DKD. Mendelian randomization (MR) analysis indicated that, among the five hub genes, only NOS3 demonstrated a statistically significant causal association with DKD. Immune infiltration analysis found that hub genes were closely related to immune cells. Molecular docking validation indicated that certain air pollutants can stably bind with hub genes such as NOS3 and PTGS2. Air pollutants may be linked to alterations in various biological processes, potentially involving key genes such as ADH5, CASP3, NOS3, PTGS2, and SDHB, including potential metabolic reprogramming, inflammatory processes, and immune microenvironment changes.
26 May 2026
Read appraisal →Medicine
Survival prediction of colorectal cancer using 101 machine learning methods based on immune-related genes: A machine learning study
Colorectal cancer (CRC) is characterized by high incidence and mortality rates. Early identification of high-risk CRC patients and timely intervention are crucial for reducing mortality rates. This study aims to develop a prognostic model incorporating immune-related genes using 101 machine learning algorithms. We utilized data from 630 CRC patients in The Cancer Genome Atlas Program database and the Gene Expression Omnibus database as a validation set. We identified immune-related genes from the ImmPort repository and conducted differential gene expression analysis. We developed 101 machine learning models, including random survival forest, elastic net, LASSO, Ridge regression, stepwise Cox, Coxboost, partial least squares regression for Cox, supervised principal component analysis, generalized boosted modeling, and survival support vector machines. The model with the highest mean concordance index was selected as the optimal prognostic model. The Ridge regression model demonstrated the most robust performance with a mean concordance index of 0.67. Patients with higher immune-related gene risk score (IRGRS) had significantly lower survival rates (log-rank test, P < .001). Univariate and multivariate Cox regression analyses confirmed IRGRS as an independent prognostic factor. Nomograms successfully predicted overall survival at 1, 2, and 3 years. IRGRS correlated with immune cell infiltration levels and stromal activity in the tumor microenvironment. High IRGRS scores were associated with higher tumor immune dysfunction and exclusion scores, suggesting potential differences in response to immune checkpoint inhibitors. In addition, high IRGRS patients exhibited greater sensitivity to certain chemotherapy drugs. The IRGRS score is a potent predictor of survival in CRC patients. This study offers a new perspective for CRC prognosis and supports the development of personalized treatment strategies.
24 May 2026
Read appraisal →Medicine
Mechanisms linking primary biliary cholangitis and osteoporosis: A combined clinical and molecular analysis
Primary biliary cholangitis (PBC) is an immune-mediated cholestatic liver disease, and osteoporosis (OP) is a prevalent comorbidity that aggravates the disease burden of PBC patients. Although the co-occurrence of the 2 diseases has been widely observed, the underlying molecular mechanisms remain unclear. Herein, we evaluated the causal link between PBC and OP using Mendelian randomization (MR) and explored shared molecular mechanisms through bioinformatics and machine learning. We performed two-sample MR using genome-wide association study summary data for PBC and OP from the Integrative Epidemiology Unit database. Inverse-variance weighting was used as the primary MR method, with heterogeneity and horizontal pleiotropy tests performed to exclude potential biases. To explore shared molecular mechanisms, we analyzed transcriptomic datasets from the Gene Expression Omnibus database, identified comorbidity-associated differentially expressed genes, and applied multiple machine learning algorithms for biomarker screening and validation, combined with immune infiltration analysis. Our study showed that MR demonstrated that PBC significantly increases the risk of OP, while transcriptomic analysis identified 36 shared differentially expressed genes enriched in key biological pathways such as ribonucleic acid splicing and ubiquitin-mediated proteolysis. Furthermore, using 3 machine learning algorithms, we identified 12 PBC-specific and 4 OP-specific diagnostic genes, whose intersection revealed vacuolar protein sorting 37 homolog C (VPS37C) as a common diagnostic biomarker. In both diseases, VPS37C exhibited an area under the receiver operating characteristic curve value >0.7, demonstrating its robust predictive performance. In addition, VPS37C expression was found to be significantly correlated with the infiltration landscape of multiple immune cell types in both PBC and OP. This study identified VPS37C as a shared diagnostic gene linking PBC and OP, providing new insights into their comorbidity at both genetic and immune levels. Our findings further elucidate the molecular mechanisms underlying the comorbidity of PBC and OP, offer novel clues for understanding their pathogenesis, and highlight promising diagnostic and therapeutic targets for clinical application.
24 May 2026
Read appraisal →Cell systems
GFETM: Genome foundation-based embedded topic model for scATAC-seq modeling
Single-cell assay for transposase-accessible chromatin with sequencing (scATAC-seq) enables investigation of open chromatin landscapes at single-cell resolution. However, analyzing scATAC-seq data remains challenging due to inherent sparsity and noise. Genome foundation models (GFMs), pre-trained on extensive DNA sequence datasets, have demonstrated effectiveness in genome analysis. Because open chromatin regions (OCRs) harbor salient sequence features, we hypothesized that leveraging GFMs' sequence embeddings could enhance scATAC-seq modeling accuracy and generalizability. We introduce the genome foundation embedded topic model (GFETM), an interpretable deep learning framework combining GFMs with the embedded topic model (ETM) for scATAC-seq analysis. By integrating DNA sequence embeddings extracted by a GFM from OCRs, GFETM demonstrates superior accuracy and generalizability and captures cell-state-specific transcription factor (TF) activity with both zero-shot inference and attention-mechanism analysis. Finally, the topic mixtures inferred by GFETM reveal biologically meaningful epigenomic signatures of kidney diabetes. A record of this paper's transparent peer review process is included in the supplemental information.
22 May 2026
Read appraisal →Genome medicine
DNA methylation biomarkers-based pan-cancer classifier: predictive modeling for cancer classification
BACKGROUND: Machine-learning (ML) driven molecular diagnostics based on omics data has a potential to revolutionize personalized medicine. However, implementation of ML into diagnostic protocols is hindered by methodological challenges which often lead to inflated performance assessment of models during development followed by poor performance of these models in implementation phase. Here, we aimed to develop and validate a pan-cancer classification framework based on DNA methylation data, that addresses methodological challenges of omics data powered ML. METHODS: We curated a primary dataset of DNA methylation profiles for 10756 samples, that included 54 healthy and cancer tissue types and validation dataset comprising data for 2306 samples from 28 independent studies. The classification framework was build using custom biomarkers selection strategy based on effect size metric that considers variance and class imbalance. The ML models were trained, tuned and evaluated using nested cross-validation approach. Local outlier factor algorithm was built into the inference pipelines to identify and filter samples displaying technical or biological anomalies. Additionally, for methodological validation of our framework we used methylation profiles for 3905 central nervous system (CNS) tumors. RESULTS: We found that relatively simple ML models outperformed complex algorithms such as deep neural network. A logistic regression classifier achieved a balanced accuracy (BACC) of 0.90 to classify 54 cancer and healthy tissue types using methylation levels at 1208 CpG sites. Similarly, our CNS tumor classifier also based on logistic regression algorithm reached a BACC of 0.94 across 59 CNS tumor subtypes. The anomaly filtering improved performance across all categories of samples tested. CONCLUSIONS: Our study demonstrates that DNA methylation profiling, when combined with carefully controlled ML practices allows for development of robust solutions that might substantially increase the efficacy of oncological diagnosis. Finally, we deployed our inference pipelines for public access via secure web platform - https://opp.pum.edu.pl/ .
21 May 2026
Read appraisal →Cell reports methods
A tri-modal contrastive learning framework for protein representation learning
Protein foundation models, particularly protein language models, have shown strong success in learning meaningful protein representations using transformer architectures pretrained on large-scale datasets through self-supervised learning. These representations have proven effective for downstream tasks such as predicting protein functions and properties. However, most existing models focus solely on amino acid sequences, overlooking other informative modalities such as 3D structures and literature text. While some recent efforts incorporate multiple modalities, they often suffer from limitations in modality coverage or training strategy. To address this gap, we propose a multimodal pretraining framework that integrates three complementary modalities-protein sequences, structures, and literature text. Our method uses the sequence modality as an anchor and aligns the other two modalities to it via contrastive learning, enabling the model to capture richer and more holistic protein representations. Across a diverse set of downstream tasks, ProteinAligner outperforms state-of-the-art foundation models in predicting protein functions and properties.
20 May 2026
Read appraisal →Cell reports methods
A network-based deep learning model integrating subclonal architecture for therapy response prediction in cancer
Predicting treatment response remains challenging in oncology, particularly given the growing diversity of therapeutic options. Despite efforts using gene expression signatures, or integrative multi-omics frameworks, robust and interpretable biomarkers remain limited. We present SubNetDL, a deep learning framework that integrates subclonal mutation profiles and protein-protein interaction networks via network propagation. Unlike condition-specific approaches, SubNetDL leverages somatic mutations alone and is applicable across diverse cancer types and treatment modalities. Applied to 10 TCGA cancer-drug combinations, SubNetDL achieved consistently strong performance (median area under the receiver operating characteristic curve [AUROC] = 0.74) and successfully generalized to two independent immunotherapy datasets (median AUROC = 0.77). Importantly, it identified candidate biomarker genes with treatment-specific relevance. SubNetDL prioritized genes that were not central in the network, highlighting its ability to capture context-specific patterns beyond traditional metrics. In conclusion, our approach offers a robust and interpretable framework for identifying predictive biomarkers and stratifying patients based on mutation profiles and network context.
20 May 2026
Read appraisal →Computer methods and programs in biomedicine
Explainable machine learning framework for the molecular classification of triple negative breast cancer
BACKGROUND AND OBJECTIVE: The difference in molecular characteristics of Triple negative breast cancer (TNBC) aids in distinguishing between its four prominent subtypes- basal-like 1, basal-like 2, mesenchymal, and luminal androgen receptor. This study presents the first integrative framework that combines explainable AI with machine learning approaches to classify TNBC subtypes. Unlike conventional models, our approach offers interpretability while enabling biomarker prioritization by identifying key hub genes that drive subtype-specific predictions. METHODS: In the experiment 783 cases (BL1 (160), BL2 (75), M (151), LAR (106), non-TNBC (291) reported in Gene Expression Omnibus (GEO) and Genomic Data Commons (GDC) data portal were used for the analysis. The proposed framework comprises modules for the identification of gene signatures for the four-subtype followed by the classification model based on eight different machine learning algorithms. Random Forest classifier was found to be best model with 96 % testing accuracy, which was elected for Explainable framework using Shapley Additive Explanations. RESULTS: Explainable biomarker module could provide a set of 47 biomarkers which is relevant in distinguishing the four types on triple negative breast cancer. The biomarkers could have the potential to be considered for TNBC prognosis in clinical setting. CONCLUSION: Key findings highlight the hub genes CDC20, CDCA2, PIMREG, KIF2C, and CENPW, implicating pathways such as ubiquitin-proteasome signaling and microtubule dynamics. These insights pave the way for biomarker-driven therapies and precision medicine in triple negative breast cancer.
16 May 2026
Read appraisal →PloS one
Analyzing the performance of deep learning splice prediction algorithms
SpliceAI is the leading tool for predicting splice-altering variants, but restrictive licensing limits clinical adoption. While open-source implementations have been published with author-reported comparisons, independent benchmarking across diverse datasets is needed to establish equivalence. We compared the original SpliceAI with two open-source implementations (OpenSpliceAI and CI-SpliceAI) and a legacy ensemble baseline across six datasets: a curated set of 1,316 validated variants, 213 variants with splice-assay data, 99,601 variants from the SPiP splicing prediction study, 242 manually curated deep intronic pathogenic variants, and two ClinVar-derived datasets comprising 53,600 intronic variants and 58,064 variants spanning all genomic contexts. The deep learning models were also evaluated against an ensemble of four legacy splice-prediction tools. Across all datasets, the deep learning algorithms outperformed the legacy ensemble. All three deep learning algorithms showed similar performance on the larger datasets dominated by canonical splice site variants (balanced accuracies 0.889-0.977). On the deep intronic benchmark, the original SpliceAI achieved the highest balanced accuracy (0.940), outperforming both CI-SpliceAI (0.890) and OpenSpliceAI (0.841). Critically, optimal thresholds for deep intronic variants were an order of magnitude lower than standard recommendations, indicating that default thresholds would miss the majority of pathogenic deep intronic variants. A correlation analysis showed that CI-SpliceAI maintained balanced concordance across event types, whereas OpenSpliceAI showed stronger correlation for loss events than gain events. Both implementations showed high positional agreement with SpliceAI, with exact splice-site match rates exceeding 90% across event types. Together, these results demonstrate that both open-source reimplementations of SpliceAI successfully reproduce the predictive behavior of the original algorithm across multiple evaluation contexts, while consistently outperforming traditional splice prediction methods. However, performance diverges on deeply intronic variants, and standard score thresholds are poorly calibrated for this variant class regardless of algorithm choice.
15 May 2026
Read appraisal →Molecular genetics and genomics : MGG
Comparison between a conventional tool and deep learning models for RNA velocity analysis of scRNA-Seq data
Single-cell RNA sequencing (scRNA-Seq) enables analysis of gene expression at single-cell resolution. RNA velocity analysis infers the temporal dynamics of transcriptional states from the relative abundances of spliced/unspliced mRNA quantified via scRNA-Seq. Classical RNA velocity approaches, such as scVelo, implement gene-specific kinetic modeling. Deep learning methods including DeepVelo, VeloVI, LatentVelo, SymVelo, and scTour are based on variational autoencoders (VAEs), which allow to enhance the robustness and accuracy by leveraging nonlinear latent representations. Here, we systematically evaluated the performance of deep learning RNA velocity tools by comparing with the scVelo dynamical model to access the possible advantages of VAE-base methods. For this purpose, public datasets (GSE149689 and GSE203233) were initially processed using a standard scRNA-Seq pipeline. Comparisons among results of selected velocity tools were conducted using cosine similarity of velocity vectors to assess directional concordance, and by mean squared error analysis of trajectory continuity for the deep learning models. Overall, VAE methods produced significant, richer, and more directionally coherent and consistent velocity fields than the classical model. Our findings indicate that deep learning models provide more consistent and biologically plausible cell-state trajectories, although at the expense of higher computational demands and reliance on accurate splicing quantification. Altogether, our results underscore the relevance of VAE-based frameworks to advance RNA velocity analysis while highlighting the need for careful preprocessing.
14 May 2026
Read appraisal →Journal of chemical information and modeling
A Transformer for Reaction-Aware Compound Explorations with GFlowNet in QSAR-Guided Molecular Design
In drug discovery tasks, achieving a balance between high biological activity toward therapeutic targets and synthetic chemical feasibility is critically important. While the recently proposed deep learning-based molecular generation models have enabled explorations of vast chemical spaces, most existing approaches do not consider synthetic routes for generated compounds. To address this issue, TRACE-GFN is proposed for molecular optimization; this method incorporates chemical reaction pathways into a quantitative structure-activity relationship (QSAR)-guided molecular design procedure. The method integrates a transformer model to explicitly learn chemical reactions with a generative flow network (GFlowNet) that efficiently samples diverse candidates. In benchmark experiments involving dopamine receptor D2 (DRD2), AKT serine/threonine kinase 1 (AKT1), and C-X-C motif chemokine receptor 4 (CXCR4), TRACE-GFN demonstrated the ability to identify compounds with high QSAR values while maintaining strong diversity, outperforming the existing molecular generation models. These results demonstrate that the proposed model can efficiently explore promising compounds while accounting for real-world chemical reactions. The source code is publicly available under an MIT license at https://github.com/sekijima-lab/TRACE-GFN.
13 May 2026
Read appraisal →Journal of chemical information and modeling
mDADGAN: Predicting miRNA-Drug Response Associations Using lncRNAs and a Diffusion-Based Generative Adversarial Network
MicroRNAs (miRNAs) are key regulators of anticancer drug response, including sensitivity and resistance. Accurate prediction of miRNA-drug response associations is important for understanding response mechanisms and guiding targeted therapy. Existing prediction methods are difficult to fully capture complex high-order biological interactions and are sensitive to data sparsity and class imbalance. In this paper, we present an miRNA-drug response association prediction method called mDADGAN by constructing a diffusion-based generative adversarial network. First, we curated a large-scale and biologically informative data set from four public data resources by aggregating functional associations to facilitate model training. Second, we built a miRNA-lncRNA-drug interaction network by integrating lncRNA into the miRNA-drug network to address data sparsity and employed heterogeneous graph convolution to extract latent features from the interactions among miRNAs, lncRNAs, and drugs. Third, we applied diffusion-based perturbation to denoise and enhance the robustness of feature representations. Finally, we performed cost-sensitive learning to optimize the objective function in an adversarial training process to address class imbalance. The experimental results on four data sets demonstrated that, compared with the existing six methods, mDADGAN achieved higher values for AUC, AUPR, ACC, Recall, and F1-score. The case study demonstrates that integrating lncRNA information effectively enhances miRNA-drug response association prediction. mDADGAN can not only accurately identify known associations in the data sets but also predict previously unreported candidate associations supported by published evidence. Overall, our method mDADGAN has the ability to identify potential miRNA-drug response associations and can provide guidance for future biological experiments to enhance experimental efficiency and reduce experimental costs.
13 May 2026
Read appraisal →Journal of chemical information and modeling
Examining Genetic Variants Associated with FOXP1 Syndrome through Molecular Dynamics of Its DNA-Binding Domain and Self-Organizing Maps
Genetic mutations in the transcription factor FOXP1 (forkhead box protein P1) cause an autosomal dominant neurodevelopmental disorder called FOXP1 syndrome. To understand the structural impact of pathogenic variants associated with FOXP1 syndrome, we investigated the conformational changes resulting from six distinct missense variants in FOXP1 by combining molecular dynamics simulations, molecular docking, and machine learning via self-organizing maps. Our results reveal different conformational landscapes mapped by the FOXP1 variants and reduced interactions with the DNA for mutations residing in helix H3 of the DNA-binding domain. These analyses offer a framework for assessing the structural impact of missense variants implicated in the FOXP1 syndrome, highlighting the importance of structural inferences in interpreting genetic variants.
13 May 2026
Read appraisal →Journal of chemical information and modeling
scFNSA: A Factorized Node-Set Attentive Framework for Single-Cell Multi-Omics Integration
Single-cell multi-omics technologies enable simultaneous interrogation of transcriptional and epigenomic states at single-cell resolution, providing powerful means to dissect cellular heterogeneity and regulatory mechanisms. However, effective integration of single-cell multi-omics data remains challenging due to extreme sparsity, high dimensionality, and cross-modal heterogeneity, which often lead to distorted similarity structures, information loss, and limited biological interpretability in existing methods. Here, we propose scFNSA, a factorized node-set attentive framework for integrative analysis of single-cell multi-omics data that explicitly decouples modality-specific feature learning from graph-based relational modeling. scFNSA first learns aligned low-dimensional representations for individual modalities through multi-view variational inference and subsequently models complex cellular dependencies via attention-guided graph learning with structured node masking. This design enables robust cross-modal structural alignment while effectively mitigating noise propagation and oversmoothing commonly encountered in deep graph neural networks. Across multiple single-cell multi-omics datasets, scFNSA consistently improves integrative representation quality and cell type resolution compared with state-of-the-art approaches. By decoupling feature extraction from relational inference, scFNSA provides a robust and interpretable framework for single-cell multi-omics integration, facilitating more accurate characterization of cellular states and underlying regulatory landscapes.
13 May 2026
Read appraisal →Journal of chemical information and modeling
DRHIN: An Integrated and Interactive Web Server for Drug Repositioning
Drug repositioning (DR) identifies new therapeutic uses for approved drugs, reducing development burdens and offering safer treatment options for patients. While high-throughput technologies generate complex, large-scale multiomics data, existing DR tools struggle to comprehensively analyze the resulting biological networks. To address this challenge, we present DRHIN, an integrated, interactive web server for DR over heterogeneous information networks (HINs) using advanced deep learning techniques. DRHIN integrates transcriptomics, proteomics, and microbiome data, incorporating eight biological entities and 19 association types to build diverse HINs and elucidate the underlying molecular mechanisms. It includes 19 state-of-the-art graph representation algorithms, enabling flexible training, comparison, and evaluation of heterogeneous network data. The platform provides a code-free portal supporting three key predictive tasks: discovering drug-disease associations, repurposing existing drugs for new indications, and identifying potential therapies for specific diseases, making analyses accessible and reproducible. Leveraging high-performance computing, DRHIN efficiently processes million-scale networks, ensuring practical applicability in real-world scenarios. The web server is freely accessible at http://drhin.tianshanzw.cn.
13 May 2026
Read appraisal →Immunologic research
An integrated AI-driven vaccine design process: a systematic review of workflows from generative design to translational prediction
Traditional vaccine development faced significant hurdles, including lengthy timelines and high costs, which hindered rapid responses to pathogens. Although the emergence of AI offered transformative potential, the necessity for a fully integrated workflow was often overlooked in studies focusing on individual tools. This review addressed a critical gap by synthesizing AI technologies across the vaccine design process, focusing on the integrated workflow from antigen discovery to clinical translation. A systematic framework was required to connect disparate tools and ensure seamless transitions. Consequently, this study provided a comprehensive roadmap for pandemic preparedness and vaccine discovery. A systematic analysis based on the PRISMA framework (2015-2024) was conducted, and 19 landmark articles were reviewed.It was demonstrated that the paradigm shift from predictive to generative AI offered unprecedented opportunities for developing novel antigens and adjuvants with superior immunogenicity. Synthesis of the literature revealed rapid progress toward sophisticated deep learning. Transformer models and Protein Language Models emerged as dominant for epitope prediction, while AlphaFold2 became the standard for structural modeling. The advent of generative AI for de novo antigen design represented the leading edge of the discipline. Additionally, AI-enhanced molecular dynamics and digital twin simulations accelerated clinical validation and manufacturing scalability. The "Integrated AI Workflow for Vaccine Design and Development" was emphasized as a comprehensive system and a prerequisite for sustainable innovation. Overall, this analysis served as a strategic roadmap for utilizing AI as a transformative framework for next-generation vaccine discovery and pandemic preparedness.
12 May 2026
Read appraisal →PloS one
DeepDRP: Dose-response predictions of drug pairs using deep learning based on data-driven feature representation and dose-response curve characteristics
Combination therapies have become a cornerstone of modern medicine, offering improved treatment outcomes and reduced side effects compared to monotherapies. However, the efficacy and safety of drug combinations depend heavily on the specific doses of each component, making the optimization of dosing regimens a crucial yet challenging task. Despite the importance of dose optimization, few computational methods address this challenge. Here, we propose DeepDRP, a novel approach that integrates two complementary models: the global model, which is trained using all available combinations, and the local model, which utilizes only samples with similar information to queries. The DeepDRP architecture comprises three main modules: the first module has three embedding networks to extract features from drugs, doses, and cell lines, and then predicts synergy using the fused knowledge. The second module constructs a graph using the input query and the set of similar retrieved training samples fed into a semi-supervised graph convolutional network to predict the synergy value. Finally, these two models are aggregated to have the final predicted value. To evaluate the proposed approach, it is applied to the NCI-ALMANAC dataset and O'Neil dataset. The obtained results denote that the proposed method achieves superior performance with respect to the other approaches.
11 May 2026
Read appraisal →Journal of computer-aided molecular design
NeuroPred-GMC: a dual-branch deep learning architecture for neuropeptide prediction based on gated dilated convolutional network and multi-scale convolutional network
Neuropeptides are multifunctional signaling molecules in the nervous system. By modulating synaptic transmission and integrating physiological systems, they influence a broad range of functions from pain perception to emotional regulation. Predicting neuropeptides can rapidly expand the library of potential therapeutic targets, thereby providing novel candidate molecules for drug development in areas such as analgesics, anti-anxiety medications, and weight-loss drugs. Traditional experimental methods are extremely time-consuming, labor-intensive, these promising alternative computational methods have emerged. In this study, a dual-branch deep learning architecture for neuropeptide prediction known as NeuroPred-GMC are built up based on gated dilated convolutional network with ESM-2 feature representation and multi-scale convolutional network with Prot-T5 feature representation. Dilated convolution exponentially enlarges the receptive field via increased dilation rates, gating mechanism enables dynamic, selective feature enhancement and noise suppression, and multi-scale convolution captures multi-level contextual information. On the independence test set, the accuracy of 93.24%, Sn of 93.69%, Sp of 92.79%, Pre of 92.86%, MCC of 0.8649 and the auROC of 0.9667 are obtained. The experimental results through cross-validation and independent test demonstrate that the proposed model has good robustness and generalizability, and can serve as a supplemental candidate predictor. The source datasets and codes can be freely available at https://github.com/yunyunliang88/NeuroPred-GMC .
11 May 2026
Read appraisal →Genome research
Incorporating valuable prior knowledge to improve deep learning prediction of genetic perturbation responses
Genetic perturbation response prediction plays a critical role in virtual cell research, yet the performance of current deep learning models still leaves room for improvement. In this study, we present a prior-guided response inference model (PRIM) that leverages a valuable priori knowledge of gene expression in control cells to model the effects of perturbations at the cellular level. This allows PRIM to predict the amount of change in gene expression after perturbation to better approximate the real situation. Compared with existing deep learning approaches, PRIM achieves superior performance across multiple data sets and notable advantages in predicting combinatorial perturbation responses. Moreover, it is more lightweight than current deep learning models and enables faster forward inference. Importantly, PRIM effectively captures nonadditive genetic interactions and shows the potential to uncover associations between combinatorial perturbations and new biologically meaningful phenotypes. These findings provide new insights into the application of deep learning for predicting cellular responses to genetic perturbations.
11 May 2026
Read appraisal →Analytical chemistry
FiLM-Enhanced Biologically Informed Neural Networks for Multiclass Omics Analysis and Biomarker Discovery
High-throughput proteomics enables detailed molecular phenotyping but poses challenges for predictive modeling and interpretation due to high dimensionality, sparsity, and nonlinear interactions. Biologically informed neural networks (BINNs) address these challenges by embedding pathway knowledge into network architectures, providing interpretable models of complex omics data. We present BINN-FiLM, which extends conventional BINNs by integrating feature-wise linear modulation, enabling task-specific scaling and shifting of pathway-level activations while preserving a fixed Reactome-based hierarchy. Unlike BINNs that share all pathway parameters across classes, BINN-FiLM captures class-dependent biological rewiring, making it particularly effective for multiclass classification tasks. Multiclass problems are reformulated as multitask binary learning, allowing pathway activations to adapt to disease-specific contexts. We evaluated BINN-FiLM on three multiclass proteomic data sets, and BINN-FiLM consistently outperformed conventional BINNs and standard machine learning models. FiLM modulation parameters revealed task-specific pathway activity, and SHAP-based interpretation identified key proteins and pathways driving disease discrimination.
11 May 2026
Read appraisal →Molecular pharmaceutics
mRNASyner: An Integrative Framework for Full-Length mRNA Sequence Optimization via Multimodule Synergistic Design
Messenger RNA (mRNA) technology holds great promise in biomedicine; however, its therapeutic efficacy relies heavily on translation efficiency and the in vivo stability of mRNA sequences, both of which are governed by dynamic interactions between the coding and noncoding regions. Existing optimization approaches typically focus on individual sequence components and lack the capacity for global, synergistic optimization. Moreover, their performance is often limited when processing long sequences or supporting iterative sequence refinement. To overcome these challenges, we propose a unified and interpretable in silico framework for full-length mRNA sequence design, integrating CDS optimization, UTR generation, and degradation modeling. In silico case studies based on respiratory syncytial virus (RSV) vaccine design demonstrated that mRNASyner achieved a favorable balance between translational accessibility and structural stability. mRNASyner thus enables the design of full-length mRNA sequences, supports long-sequence optimization, and offers a novel solution for the development of personalized mRNA therapeutics.
11 May 2026
Read appraisal →Briefings in bioinformatics
Comprehensive evaluation of artificial intelligence-empowered approaches for protein-aptamer complex prediction
Drug discovery is a time-consuming, expensive, and high-risk process. Recent advances in artificial intelligence (AI) have enabled major breakthroughs in small-molecule and protein therapeutics. However, AI-driven design of aptamer drugs remains largely unexplored. Aptamers are short (15-100 nt) single-stranded DNAs or RNAs that exhibit high binding affinity, high specificity, and low immunogenicity, making them promising candidates for disease (such as cancer) therapeutics. Compared with protein-ligand or protein-protein systems, protein-aptamer complexes are under-represented in public structural databases, and aptamers themselves are highly flexible and relatively large molecules. These characteristics present distinct challenges for AI-based structural modeling. Here, we systematically evaluate recent AI frameworks, including AlphaFold3, Chai-1, Boltz-2, and RoseTTAFold2NA, along with a template-based approach, in predicting protein-aptamer complex structures and estimating binding free energies. We establish an independent benchmark to assess their performance in structural accuracy, stability, and energetic consistency. This study provides a foundation for the application of AI in aptamer drug design and offers a reference framework for future research in nucleic-acid therapeutics and biomolecular modeling.
10 May 2026
Read appraisal →Briefings in bioinformatics
MLN2SVG: domain-aware spatially variable gene detection using contrastive variational autoencoder and multi-level neighbor search
Spatial transcriptomics (ST) technologies have transformed our ability to examine gene expression within intact tissues, yet accurately identifying spatially variable genes (SVGs) remains challenging due to spatial heterogeneity, data sparsity, and incomplete modeling of domain-level dependencies. To address these limitations, we propose MLN2SVG, a domain-aware framework that integrates contrastive variational autoencoding with a multi-level neighbor (MLN) search algorithm to jointly learn tissue domains and SVGs. MLN2SVG constructs a weighted spatial graph to capture both local and long-range spatial relationships, employing a deep contrastive variational autoencoder to align augmented and original data representations while preserving biological diversity. The MLN algorithm dynamically expands neighborhood connectivity to mitigate sparsity and enhance domain coherence. Across multiple human and mouse ST datasets, including dorsolateral prefrontal cortex, breast cancer, and brain tissues, MLN2SVG consistently outperformed existing methods in clustering accuracy, robustness, and biological interpretability. Notably, in breast cancer tissues, MLN2SVG uncovers fine-grained spatial organization of tertiary lymphoid structures, delineating region-specific immune architectures spanning intratumoral, tumor-edge, and extratumoral compartments. Through the integration of spatial domain discovery and SVG detection, MLN2SVG delivers a robust and biologically interpretable framework for uncovering the molecular and structural complexity of tissue organization.
10 May 2026
Read appraisal →Protein science : a publication of the Protein Society
AlphaFold-driven discovery of oxysterol-binding protein-related protein-phosphoinositide 3-, 4-, and 5-phosphatase interactions using new generation confidence scores
Non-vesicular lipid transport contributes to the regulation of membrane composition and organelle function at membrane contact sites. OSBP-related proteins (ORPs) are central to this process, yet their interaction networks remain incompletely defined. Here, we systematically screened potential interactions between ORPs and phosphoinositide 3-, 4-, and 5-phosphatases using AlphaPulldown2, AlphaFold2-Multimer, and AlphaFold3. We established a protocol for model generation by combining AlphaFold2-Multimer predictions (including five-replicates) with an AlphaPulldown2 interaction screen across around 200 protein pairs, and with AlphaFold3 predictions including lipid-bound and multimeric assemblies. Interface confidence was assessed for consistency using the weighted ipTM + pTM metric, actifpTM, new generation ipSAE scoring, and FoldSeek-Multimer clustering. We further evaluated the protein pairs' biological plausibility based on subcellular localization data, in silico membrane insertion, evolutionary conservation via ConSurf, and protein binding interface analysis using the deep learning tool PeSTo. This integrative protocol uncovered functionally conserved binding modes in the SAC1 lipid phosphatase with the ORP family, particularly with ORP11, and predicted functionally relevant protein-lipid interfaces.
9 May 2026
Read appraisal →Microscopy research and technique
Deep Learning Integration in Optical Microscopy: Advancements and Applications
Optical microscopy is a cornerstone imaging technique in biomedical research, enabling visualization of subcellular structures beyond the resolution limit of the human eye. However, conventional optical microscopy faces challenges such as optical aberrations, diffraction-limited resolution, low signal-to-noise ratio (SNR), and poor contrast. The exponential growth of bioimaging data further underscores the need for advanced computational tools. Deep learning (DL) is a subset of machine learning that has emerged as a transformative approach to address these limitations, offering enhanced precision, reduced manual intervention, and diminished reliance on domain-specific expertise for image reconstruction, enhancement, and analysis. This review explores the integration of DL into optical microscopy, focusing on key applications including image classification, segmentation, and computational reconstruction. We examine prominent DL architectures such as convolutional neural networks (CNNs), U-Nets, residual networks (ResNets), and generative adversarial networks (GANs)-and their role in advancing diverse microscopy modalities. These frameworks enhance image quality, improve quantitative analysis, and democratize access to high-performance microscopy. Additionally, we discuss persisting challenges, including the demand for large, annotated datasets, dynamic sample variability, model interpretability, and potential data biases. Collectively, DL is poised to revolutionize optical microscopy, shaping its future developments in biomedical imaging.
6 May 2026
Read appraisal →Cell
Deep-learning-based de novo discovery and design of therapeutics that reverse disease-associated transcriptional phenotypes
Identifying drugs that reverse disease-associated transcriptomic features has been widely explored for drug repurposing, but its potential for de novo drug discovery remains underexplored. Here, we present gene expression profile predictor on chemical structures (GPS), a deep-learning-based drug discovery platform, guided by transcriptomic features, that screens large compound libraries and optimizes lead molecules. We first develop a model that captures transcriptomic perturbation signatures solely from chemical structures and deploy it to library compounds. We refine scoring methods and employ a tree-search method for optimization. By incorporating structure-gene-activity relationships, we uncover drug mechanisms from transcriptomic data. We evaluate GPS across multiple diseases and conduct extensive validation in two cases. In hepatocellular carcinoma, we discover two unique compound series with favorable cellular selectivity and in vivo efficacy. In idiopathic pulmonary fibrosis, we identify one repurposing candidate and one novel anti-fibrotic compound by reversing gene expression of multiple distinct cell types derived from single-cell transcriptomics.
5 May 2026
Read appraisal →Medical image analysis
PH2ST: Prompt-guided hypergraph learning for spatial transcriptomics prediction in whole slide images
Spatial Transcriptomics (ST) reveals the spatial distribution of gene expression in tissues, offering critical insights into biological processes and disease mechanisms. However, the high cost, limited coverage, and technical complexity of current ST technologies restrict their widespread use in clinical and research settings, making obtaining high-resolution transcriptomic profiles across large tissue areas challenging. Predicting ST from H&E-stained histology images has emerged as a promising alternative to address these limitations but remains challenging due to the heterogeneous relationship between histomorphology and gene expression, which is affected by substantial variability across patients and tissue sections. In response, we propose PH2ST, a prompt-guided hypergraph learning framework, which leverages limited ST signals to guide multi-scale histological representation learning for accurate and robust spatial gene expression prediction. Extensive evaluations on two public ST datasets and multiple prompt sampling strategies simulating real-world scenarios demonstrate that PH2ST not only outperforms existing state-of-the-art methods, but also shows strong potential for practical applications such as imputing missing spots, ST super-resolution, and local-to-global prediction, highlighting its value for scalable and cost-effective spatial gene expression mapping in biomedical contexts.
5 May 2026
Read appraisal →Pakistan journal of pharmaceutical sciences
Network pharmacology and machine learning reveal multi-target mechanisms of poly herbal formulation against atherosclerosis
BACKGROUND: Herbs, like Allium sativum, Ginkgo biloba and Nerium oleander are traditional medicinal plants that have been used to treat atherosclerosis and cardiovascular disease. This study provides valuable insights into how network pharmacology (NP) and emerging machine learning are utilized to identify potential drug candidates from these plants for treatment of atherosclerosis. METHODS: NP analysis was employed to screen compounds and their potential gene targets from databases and tools e.g. IMPPAT, PubChem, KNAPSACK, Swiss ADME, Swiss Target Prediction, Disgenet and GeneCard. Cytoscape 3.10.2 was employed to visually understand these networks. DAVID database was used for functional and enrichment analysis of the genes validated through molecular docking using PyRx and Discovery Studio. RESULTS: Computational tools and bioinformatics approaches showed a few core compounds, such as Quercetin, Naringenin, Luteolin, Kaempferol, Apigenin, Daidzein, Luteolin-7-olate, Pinocembrin, Pregnenolone and Fisetin found to be effective against atherosclerosis. Pathway analysis revealed that mechanism of atherosclerosis development is directly associated with cholesterol metabolism, cellular senescence, Ras, NF-κB and PI3K-Akt signaling pathways. CONCLUSION: NP and molecular docking analysis suggested that screened compounds may inhibit progression of atherosclerosis by modulating key associated pathways. Hence, this machine learning aided NP study provides basis for understanding and recognizing the activity of these plants in treating atherosclerosis.
4 May 2026
Read appraisal →Journal of molecular graphics & modelling
DeepHybridCPI: A hybrid deep learning framework for compound-protein interaction prediction
In bioinformatics, deep learning-based methods for Compound-Protein Interaction (CPI) prediction play a vital role in virtual screening, drug discovery, and drug repositioning. Recent improvements in computational methods have shown great possibility to save costs of experiment and speed up target identification. Nevertheless, the current CPI forecasting methods remain severely limited. Many rely on shallow Graph Neural Networks (GNNs) that struggle to capture the global structural context of compounds, while conventional Convolutional Neural Networks (CNNs) focus primarily on local sequence motifs and fail to model long-range dependencies in proteins. Even though a number of recent architectures strive to solve these problems by adding complexity to models, or by adding complex modules, these additions often cause significant computational overhead. To overcome these challenges, we propose DeepHybridCPI, a hybrid deep learning framework designed for accurate and efficient CPI prediction. Our hybrid model integrates a multiscale, densely connected GNN to extract compound features capturing both local substructures and global molecular topology, and employs CNNs with Long Short-Term Memory (LSTM) networks to model both local motifs and extended dependencies in protein sequences. The learned compound and protein representations are fused into a unified latent space to enable effective interaction modeling. Experimental evaluations on benchmark Human and C. elegans datasets demonstrate that DeepHybridCPI consistently outperforms existing state-of-the-art baseline methods in terms of AUC, Precision, and Recall. These findings highlight the importance of combining multiscale compound representations with hybrid sequence encoders within a single unified framework, providing a promising avenue for accelerating computational drug discovery. We release our source code and dataset at: https://github.com/jamshaidwarraich/DeepHybridCPI.
4 May 2026
Read appraisal →Computer methods in biomechanics and biomedical engineering
Multiple instance learning method based on convolutional neural network and self-attention for early cancer detection.
Early cancer detection using T-cell receptor sequencing (TCR-seq) and multiple instances learning methods has shown significant effectiveness. We introduce a multiple instance learning method based on convolutional neural networks and self-attention (MICA). First, MICA preprocesses TCR-seq using word vectors and then extracts features using convolutional neural networks. Second, MICA uses an enhanced self-attention mechanism to extract relational features of instances. Finally, MICA can extract the crucial TCR-seq. After cross-validation, MICA achieves an area under the curve (AUC) of 0.911 and 0.946 on the lung and thyroid cancer datasets, which are 7.1% and 2.1% higher than other methods, respectively.
4 May 2026
Read appraisal →Biotechnology advances
Computational methods for signal peptide prediction: From statistical models to deep learning
Signal peptides are short amino acid sequences located at the N-terminus of proteins. They guide newly synthesized proteins to their correct cellular destinations, playing a crucial role in protein localization and transport. Traditional experimental methods for identifying signal peptides are typically time-consuming, costly, and labor-intensive, driving rapid development of computational alternatives. Over the past two decades, researchers have proposed various computational approaches, with prediction accuracy continuously improving through evolution from early statistical and rule-based algorithms to deep learning. In this review, we systematically summarize these computational approaches, emphasizing methodological evolution and framework design. We compile representative computational methods, comparing their prediction outcomes and identifying existing limitations. Finally, we discuss current challenges and emerging opportunities, aiming to advance the development of computational frameworks characterized by unified evaluation, biologically grounded interpretation, and generative modeling.
4 May 2026
Read appraisal →Molecular informatics
ADME-DTI: Augmented Deep Meta Ensemble for Drug-Target Interaction Prediction
Drug-target interaction represents a critical focus area in computational drug discovery and pharmaceutical research. However, the process of identifying and analyzing these interactions is often resource-intensive, requiring extensive experimentation to evaluate the binding relationships between numerous drugs and their respective targets. This complexity is further compounded by the fact that a drug can inhibit multiple targets, and a target may also bind to various drugs. To address these issues, advanced deep learning models have been introduced as promising tools, offering the ability to accurately predict binding affinity and other bioactivity values to distinguish between potential drug-target interactions. The proposed model, named the Augmented Deep Meta Ensemble for Drug-Target Interaction (ADME-DTI), leverages multiple descriptors and fingerprint representations to extract meaningful insights from drug and protein data. These submodels from each representation are then combined with a deep learning architecture along with the metadata of the drug and target entries. The proposed approach has proved competitive performance against state-of-the-art models across diverse datasets and evaluation metrics, as evidenced by key metrics, namely r m 2 $r_m^2$ , concordance index, and mean squared error. Specifically, the model achieved mean squared error values of 0.186 (Davis), 0.118 (Kiba), 0.134 (DTC), 0.262 (Metz), 0.300 (ToxCast), and 0.791 (STITCH), all of which are publicly available benchmark datasets commonly used in drug-target interaction prediction tasks. The results highlight the model's effectiveness in reducing prediction errors and improving accuracy. Previous drug-target interaction research often relied on limited, non-diverse datasets, reducing generalizability. Few studies addressed drug-target interaction prediction as a regression task. Our novel deep learning metamodel integrates multiple models and representations, surpassing benchmarks and delivering greater prediction accuracy across varied datasets and evaluation metrics.
4 May 2026
Read appraisal →Nano letters
Seeing and Feeling DNA Methylation: Single-Molecule Biophysics Meets Machine Learning
DNA methylation at 5-methylcytosine (5mC) is crucial for embryonic development and cellular function, while aberrant patterns strongly drive disease onset and progression. Its reversible nature offers substantial therapeutic potential, emphasizing the need for precise, context-specific genome wide 5mC mapping. Conventional techniques such as bisulfite sequencing and ensemble biosensor assays are hindered by DNA degradation, amplification bias, high cost, and inability to resolve single-molecule structural and mechanical effects of methylation. This review examines advances in single-molecule biophysical methods (nanopore sensing, smFRET, optical/magnetic tweezers, and AFM) that provide direct, label-free/minimally invasive 5mC detection, along with quantitative insights into DNA conformation, mechanics, and protein-DNA interactions. These techniques complement traditional methylome mapping by linking genomic localization to molecular mechanisms. Emerging machine-learning approaches are revolutionizing analysis, particularly in nanopore sensing, while promising applications in smFRET, tweezers, and AFM address throughput and reproducibility challenges. Their convergence promises scalable, high-resolution epigenetic profiling, advancing precision epigenomics toward clinical application.
4 May 2026
Read appraisal →Scientific reports
scDecorr: feature decorrelation based representation learning enables self-supervised alignment of multiple single-cell experiments
Single-cell RNA sequencing (scRNA-seq) has revolutionized our understanding of cellular heterogeneity in complex biological systems. However, analyzing and integrating scRNA-seq data poses unique computational challenges due to sparsity, high variability, and technical batch effects. Here, we propose a novel framework called scDecorr for robust representation learning and data integration for scRNA-seq analysis. Our approach leverages the idea of feature decorrelation-based self-supervised learning (SSL) to obtain efficient low-dimensional representations of individual cells without relying on cell-type annotations. By maximizing similarity among distorted embeddings while decorrelating their components, scDecorr captures the biological signature while eliminating technical noise. Furthermore, scDecorr incorporates unsupervised domain adaptation to bridge the gap between batches with different distributions, enabling effective integration of scRNA-seq data from diverse sources. Our framework achieves domain-invariant representations by learning cell embeddings independently across domains and employing domain-specific batch normalization. We evaluate scDecorr on a variety of single-cell datasets and demonstrate its ability to integrate batches without losing the inherent biological variance, thereby facilitating optimal clustering. The representations generated by scDecorr also exhibit robustness in label transfer tasks, allowing for effective transfer of cell-type labels from reference to query datasets. Overall, scDecorr offers a powerful tool for efficient analysis and integration of large and complex scRNA-seq datasets, advancing our understanding of cellular processes and disease mechanisms. The code is available here https://github.com/hayatlab/scdecorr .
3 May 2026
Read appraisal →PLoS computational biology
MsgaBpred: A B-cell epitope predictor integrating AlphaFold3-predicted structures with multi-scale GCNs and pre-trained language model ESM-C
Accurate prediction of B-cell epitopes plays a key role in facilitating advancements in vaccines, therapeutics, and diagnostics. In contrast to labor-intensive experimental approaches, computational strategies provide a more economical and efficient means of identifying potential epitopes. Existing methods are often limited by their reliance on experimentally resolved protein structures or by the use of lower-accuracy predicted structures. Sequence-based approaches, while fast, largely fail to capture the 3D spatial context essential for conformational epitopes. With the breakthroughs achieved by AlphaFold3 in predicting protein structures, we present MsgaBpred, the model to apply AlphaFold3-derived structures to B-cell epitope identification. Given only a protein sequence, our model employs a multi-scale graph convolutional network and additive attention to capture complex structural dependencies without relying on experimentally determined structures. The multi-scale design allows for effective modeling of both local and global contexts by aggregating information across different neighborhood ranges. Additionally, we leverage ESM-C, a more expressive protein language model than ESM-2, to enhance feature representation for B-cell epitope prediction. Extensive evaluations across multiple benchmark datasets demonstrate that MsgaBpred achieves competitive and robust performance; notably, it yields a statistically significant improvement in AUC compared to existing state-of-the-art methods. Moreover, the modular and scalable architecture of MsgaBpred holds promise for broader applications, including the structural analysis of other biomolecular entities such as nucleic acids and carbohydrates.
3 May 2026
Read appraisal →Journal of chemical information and modeling
TriPDCL: A Tri-Pathway Prototype-Driven Contrastive Learning Framework for Cross-Modality Single-Cell Integration
The rapid development of single-cell sequencing technologies has laid a solid foundation for multiomics data integration. A central objective in multiomics integration is to attenuate cross-omic discrepancies stemming from technical batch effects and inherent biological variability, thereby facilitating a unified latent representation that supports robust downstream analyses. However, the intrinsic high sparsity of single-cell data and pronounced intercellular heterogeneity pose greater challenges for joint analysis. We foreground cellular heterogeneity and present TriPDCL, a prototype-based contrastive learning framework, which not only achieves effective transfer of heterogeneity information through an iterative prototype-learning update mechanism but also, by leveraging learnable prototype centers, enables precise construction of reliable positive-negative sample pairs. It tackles two principal challenges, the alignment of cross-modal heterogeneity and the robust learning, in multiomics analysis. Subsequently, comparative assessments conducted on five data sets in contrast to seven representative methods illustrate the superiority of the proposed approach.
2 May 2026
Read appraisal →Journal of chemical information and modeling
DWPL-GCNMF: Structure-Aware Dynamic Weighted Pseudo-Label Learning for Adverse Drug Reaction Prediction
Adverse drug reaction (ADR) prediction is typically formulated as drug-ADR association learning on extremely sparse, positive-unlabeled matrices, which limits generalization and makes naive pseudolabel expansion prone to confirmation bias. Here, we propose DWPL-GCNMF, a structure-aware semisupervised framework that combines graph convolutional network (GCN)-based structural embeddings learned from a drug-protein knowledge graph with matrix factorization scoring and introduces a prediction-consistency-driven dynamic weighted pseudolabel learning strategy to expand training pairs while down-weighting unreliable pseudolabels; multiple base models are further fused to improve prediction stability. We evaluate DWPL-GCNMF on a DrugBank-derived benchmark (1,177 drugs × 4,247 ADRs; ∼2.6% observed associations) and an external SIDER benchmark (1,080 drugs × 5,579 ADRs; ∼2.3% observed associations) to assess robustness under different data sets and increased sparsity. DWPL-GCNMF consistently outperforms representative baselines, achieving AUPR/F1 of 0.6656/0.6194 on DrugBank and 0.6106/0.5816 on SIDER, while also improving top-K prioritization (e.g., Recall@15 up to 0.8078 and 0.8146, respectively), indicating that structure-aware representation learning with dynamically weighted pseudolabeling provides a robust approach for prioritizing potential ADRs under severe data sparsity.
2 May 2026
Read appraisal →Journal of chemical information and modeling
Explainability Methods from Machine Learning Detect Important Drugs' Atoms in Drug-Target Interactions
Predicting drug-target interactions (DTI) with graph neural networks (GNNs) is hindered by their lack of interpretability. To address this, we benchmark four explainable artificial intelligence (XAI) attribution methods on GNN models trained for kinase and G-protein-coupled receptors (GPCR) targets. We assess the methods' consistency through atom-level intersection over union (IoU) and validate their biological relevance by mapping attributed atoms to three-dimensional (3D) protein-ligand structures. While consistency across methods was modest, consensus attributions were highly enriched for atoms directly contacting the binding pocket─up to 76% within 2 Å in the kinase-inhibitor complexes. Notably, these attributed atoms were frequently found contacting experimentally important regulatory residues such as those in the DFG motif. This indicates that XAI methods, despite their disagreements, can identify chemically meaningful ligand features, providing a foundation for developing more interpretable GNNs in drug discovery.
2 May 2026
Read appraisal →Journal of chemical information and modeling
A Reinforcement Learning-Guided Genetic Algorithm Integrating Medicinal Chemistry-Inspired Molecular Transformations
Achieving optimal target activity while maintaining synthetic accessibility and drug-likeness represents a major challenge in computational drug discovery. Existing de novo generative models often yield chemically invalid or synthetically intractable structures and struggle to optimize multiple objectives simultaneously. Here, we introduce ALCHIMIA, an interpretable hybrid framework combining reinforcement learning (RL) and a genetic algorithm (GA), built based on a vocabulary of 33 medicinal chemistry-inspired molecular transformations. The RL component trains a policy network to prioritize transformation sequences that improve synthetic accessibility (SA) and the quantitative estimate of drug-likeness (QED) scores, embedding these constraints directly into molecular generation. The GA component applies the learned policy as a mutational operator within population-based optimization guided by molecular docking, enabling the exploration of diverse chemical lineages while converging toward high-affinity ligands. ALCHIMIA was applied to two different pharmacologically relevant targets: human Cannabinoid Receptor 2 (CB2R) and human Sigma nonopioid intracellular Receptor 1 (S1R). We considered three different scenarios: (i) unconstrained hit identification; (ii) scaffold-constrained lead optimization; and (iii) design of dual modulators. The framework generated chemically valid molecules with QED and SA scores comparable to or better than those obtained with random baselines and selected de novo design methods. By codifying typical medicinal chemistry actions as learnable transformations and coupling multiobjective optimization with GA-based diversity maintenance, ALCHIMIA, freely available as a GitHub repository (https://github.com/alberdom88/ALCHIMIA), provides a practical, interpretable, and scalable framework for molecular de novo design.
2 May 2026
Read appraisal →Scientific reports
LASSO-HHO two-stage hybrid gene selection framework for accurate Alzheimer's disease diagnosis
Selecting informative genes is essential for building accurate and efficient diagnostic models, especially for high-dimensional gene expression data such as those related to Alzheimer's disease. These datasets typically involve thousands of genes with limited samples, increasing the risk of overfitting and computational complexity. To address this, we propose a LASSO-HHO Gene Selection (LHGS) framework. LASSO is first applied to reduce dimensionality by filtering irrelevant genes. A conditional HHO-based optimization stage is then applied only when the LASSO-selected subset does not achieve sufficient accuracy or remains relatively large. Otherwise, the LASSO-selected features are directly used without further optimization. Experimental results show that the proposed method reduces the number of selected genes by up to 99.9% while maintaining high performance. The experimental results indicate that 100% accuracy can be achieved on specific datasets. In particular, GSE48350 and GSE36980 achieved 100% accuracy using LASSO alone, whereas GSE118553 and GSE132903 required the full LHGS framework to achieve the same performance. The framework also improves computational efficiency within a consistent experimental setup by reducing the optimization search space after LASSO filtering. Overall, LHGS provides a practical and efficient solution for gene selection in high-dimensional biomedical data.
28 Apr 2026
Read appraisal →Journal of ethnopharmacology
MAGED: Multimodal attentive graph learning with gene expression dynamics on knowledge graphs for TCM target prediction
ETHNOPHARMACOLOGICAL RELEVANCE: Plants used in traditional medicine have long provided valuable clues for drug discovery, yet systematically connecting their documented therapeutic effects with modern understanding of biological mechanisms remains a key scientific question. Bridging the gap between traditional knowledge, clinical symptom profiles, and contemporary molecular-level regulation is essential for advancing the study of herbal medicine. OBJECTIVES: This study aims to develop a multimodal graph attention learning framework, named MAGED, which integrates knowledge graphs with gene expression dynamics to improve the accuracy and interpretability of Herb-Target Interaction (HTI) prediction. METHODS: MAGED incorporates multi-source heterogeneous data, including traditional Chinese medicine (TCM) properties, hierarchical clinical symptom information, and molecular-level biological data, such as gene expression dynamics under drug intervention and causal biological networks. A multimodal fusion encoder is employed to embed herb attributes into efficacy-related feature vectors, which are then combined with functional representations (e.g., normalized enrichment scores) to form a unified contextualized representation. Additionally, a hierarchical graph attention network is designed to integrate macroscopic symptom-gene associations with microscopic regulatory pathways, thereby establishing a coherent inference chain from herbal effects to biological targets and phenotypic symptoms. RESULTS: Systematic experimental evaluations demonstrated that MAGED significantly outperforms existing baseline methods across multiple evaluation metrics, achieving a 59.8 % improvement in HR@10. The model also exhibited superior ranking and recall performance under cold-start scenarios. In a case study on Scutellaria baicalensis, 8 out of the top 10 predicted targets were supported by existing literature, with functional or direct interaction evidence, and some targets were further validated through molecular docking. CONCLUSION: MAGED provides an accurate and interpretable framework for predicting herb-target interactions, effectively integrating traditional knowledge with modern molecular evidence. This approach shows strong potential to facilitate the discovery of herbal mechanisms and the identification of novel therapeutic targets.
27 Apr 2026
Read appraisal →Proceedings of the National Academy of Sciences of the United States of America
Machine learning multiscale collective cell dynamics: From single-cell characterization to multicellular monolayer modeling
Collective cell dynamics are fundamental to numerous physiological processes, including embryo development, tissue morphogenesis, immune response, and disease progression. Accurately modeling these dynamics across scales remains challenging, as traditional physics-based models usually rely on many unmeasurable parameters and unclear active physics, limiting their ability to capture both single-cell features and emergent multicellular behaviors. Here, we introduce a scale-adaptive hybrid machine learning (ML) framework, in which complementary physics-guided and physics-agnostic method each resolves an above key issue at single-cell and multicellular scales, respectively. The physics-guided ML method integrates physical models with experimental data to infer previously unmeasurable parameters, enabling more accurate characterization of single-cell shape and velocity features, but unknown active forces and inherent physical assumptions still constrain its ability to predict multicellular scale dynamics. In contrast, the physics-agnostic ML method bypasses explicit physical assumptions to directly model multicellular behaviors from historical state sequences. This method provides robust long-term predictions of coarse-grained density oscillations and waves observed experimentally, yet it suffers from high stochasticity when applied to single-cell scale modeling. By assigning each approach to the scale where it is most effective, our framework leverages their respective strengths while mitigating their limitations. This complementary strategy bridges the gap between theoretical modeling and experimental observations, offering a versatile computational paradigm for multiscale collective cell dynamics with broad potential in diverse physiological and pathological contexts.
27 Apr 2026
Read appraisal →Cell reports methods
A high-speed attention network for MHC-bound peptide identification and 3D modeling
We developed SwiftMHC, an ultra-fast and accurate structure-based framework for peptide-MHC (pMHC) modeling and binding affinity prediction. Using task-specific deep learning trained on physics-derived synthetic data, SwiftMHC predicts pMHC binding affinities in 0.009 s per case on a single A100 GPU when running in batch mode, offering improved speed compared with leading sequence-based tools such as netMHCpan and MHCflurry while maintaining competitive accuracy. In addition, SwiftMHC generates all-atom 3D pMHC structures with a median Cα-RMSD of 1.32 Å against crystallographic benchmarks, matching or exceeding state-of-the-art methods such as AlphaFold2-finetune but at a lower computational cost. Optimized for HLA-A∗02:01 9-mer peptides but readily extensible to other alleles, SwiftMHC unites structural insight with high-throughput scalability to accelerate safe and effective epitope discovery in cancer immunotherapy.
24 Apr 2026
Read appraisal →Analytical chemistry
scAClc: A Multi-Objective Adaptive Clustering Framework for Single-Cell Transcriptomics via Contrastive and Resolution-Aware Representation Learning
Single-cell RNA sequencing (scRNA-seq) enables whole-transcriptomic profiling at single-cell resolution, facilitating the construction of virtual cell representations that capture the full spectrum of cellular identities. Realizing this goal hinges on accurate clustering, which remains challenging due to data sparsity, high dimensionality, noise, and the need to specify cluster numbers a priori. We propose scAClc, a novel clustering framework featuring multiobjective optimization and adaptive resolution discovery, designed to address these limitations through three key innovations. First, a Hierarchical Gene Relevance Module integrates global gene variability with local neighborhood-specific signals to eliminate redundancy while retaining biologically informative features. Second, an Anchor-Centered Contrastive Learning Module adaptively selects representative anchors to guide embedding learning, promoting compact intracluster structure and clear intercluster separation. Third, based on the robust low-dimensional embedding, we propose a Self-Adaptive Resolution Discovery Module to automatically infer the number of clusters by jointly modeling intra- and intercluster distances. Extensive experiments on 15 real scRNA-seq data sets demonstrate that scAClc consistently outperforms six state-of-the-art methods across multiple evaluation metrics. Ablation studies further confirmed the complementary contributions of each module. In addition, interpretability analysis effectively mitigates the "black box" nature of clustering models and sheds light on the biological mechanisms underlying cell clustering. The source code is publicly available at https://github.com/scAClc/scAClc.
23 Apr 2026
Read appraisal →Journal of medical genetics
Clinically significant DNA variation from the GENCOV and HostSeq COVID-19 genome sequencing studies
BACKGROUND: Vast amounts of genome sequencing data generated from large-scale research studies like HostSeq provide an opportunity to summarise the spectrum of pathogenic variation in a subset of the Canadian population. Sharing variant-level data with public databases, like ClinVar, is crucial for advancing our understanding of genomic variants related to Mendelian diseases. However, such entries are often incomplete or contain discrepancies which render interpretation and classifications less useful. METHODS: The GENCOV and HostSeq cohorts were annotated and summarised using custom workflows to identify variants with a pathogenic and likely pathogenic (P/LP) classification in ClinVar. These variants were further filtered using custom gene panels, gnomAD frequency, variant type, gene-disease relationship and the number of reputable ClinVar laboratory submissions. Manually assessed variants from the GENCOV study were compared with ClinVar classifications to identify discrepancies. RESULTS: A total of 1956 unique variants were manually classified as P/LP through the GENCOV study. Of these, 65% (1276) were also identified in ClinVar, and among those, 69% (889) had concordant P/LP classifications. The manual assessment of unique exonic variants in GENCOV yielded a higher number of putative P/LP variants (1595), including truncating and missense variants compared with the 127 unique exonic P/LP variants in HostSeq. CONCLUSION: These results highlight that a large proportion of P/LP variation is either absent or has conflicting evidence for pathogenicity in ClinVar, emphasising the importance of periodic reassessment, discrepancy resolution and updates to ensure completeness and accuracy of public databases.
23 Apr 2026
Read appraisal →Biological cybernetics
Geometric Learning Dynamics
We present a unified geometric framework for modeling learning dynamics in physical, biological, and machine learning systems. The theory reveals three fundamental regimes, each emerging from the power-law relationship g ∝ κ α between the metric tensor g in the space of trainable variables and the noise covariance matrix κ . The quantum regime corresponds to α = 1 and describes Schrödinger-like dynamics that emerges from a discrete shift symmetry. The efficient learning regime corresponds to α = 1 2 and describes very fast machine learning algorithms. The equilibration regime corresponds to α = 0 and describes classical models of biological evolution. We argue that the emergence of the intermediate regime α = 1 2 is a key mechanism underlying the emergence of biological complexity.
20 Apr 2026
Read appraisal →mAbs
Multi-modal feature learning to prioritize ADCs with favorable half-life in mice
Favorable pharmacokinetic (PK) properties of antibody-drug conjugates (ADCs) enable higher systemic exposure and are associated with improved clinical efficacy, yet ADC PK optimization remains largely empirical and resource-intensive. We developed an in silico framework to predict half-lives of monoclonal antibodies (mAbs) and ADCs in C57BL/6 mice using multi-modal feature learning. To address structural complexity of ADCs, each compound was deconvoluted into three components: (1) antibody features represented using sequence descriptors, AlphaFold-based structural properties, or protein language model (PLM) embeddings; (2) conjugation-related features, including drug-to-antibody ratio and conjugation sites; and (3) physicochemical properties of linker-payloads. XGBoost models were trained using experimental half-life data from 118 mAbs and ADCs. Structure-based and PLM approaches achieved strong cross-validation (Pearson's r = 0.76), outperforming sequence-only models. In temporal validation of 32 mAbs and ADCs, the structure-based models demonstrated robust generalization (r = 0.79). PLM embeddings achieved comparable rank ordering (Spearman's ρ up to 0.81 for ESM-2), but lower quantitative accuracy relative to structure-based models. Overall, our study presents a scalable machine learning framework for early PK screening of mAbs and ADCs prior to synthesis and in vivo testing.
19 Apr 2026
Read appraisal →Cell systems
Context-informed subgraph foundation models enable interpretable protein-function prediction
Protein-function prediction is crucial for elucidating molecular mechanisms driving biological processes and therapeutics development. Despite numerous computational tools demonstrating promising performance, they fall short when predicting rare, uncharacterized functions or indirect activities. Here, we present COSMOS, a context-aware Gene Ontology (GO) subgraph mining system for protein-function prediction. By leveraging inductive subgraph foundation models and an enriched knowledge graph of protein-GO relationships, COSMOS performs zero-shot, few-shot, and low-homology protein-function prediction. Built on 7,923,952 functional semantic relationships, COSMOS demonstrates robust capabilities to (1) generate state-of-the-art predictions for GO classes with sparse or no experimental annotations, (2) provide interpretable functional subgraphs for transparent rationale analysis, and (3) deliver complementary benefits when integrated with existing embedding-based prediction methods. We anticipate that COSMOS will serve as a complementary approach to conventional protein annotation methods and an interpretable tool for predicting protein functions within underexplored GO classes, thereby advancing genomics and therapeutic research.
18 Apr 2026
Read appraisal →PloS one
upsML: A high-accuracy machine learning classifier for predicting Plasmodium falciparum var gene upstream groups
Plasmodium falciparum erythrocyte membrane protein 1 (PfEMP1), encoded by the hypervariable var gene family, is central to malaria pathogenesis, influencing both disease severity and immune evasion. Classifying var genes into upstream groups (upsA, upsB, upsC, upsE) is important for understanding parasite biology and clinical outcomes, but remains challenging, especially with partial sequences, such as the DBLα tag or RNA-Seq assemblies. We developed upsML, a machine-learning-based classifier trained on 2,530 curated var genes, to accurately assign upstream groups based on sequence features from different partial gene regions. We compared seven methods, including support vector machines, random forests, XGBoost, and HMMER models. Several models in upsML achieve accuracies of 83% for DBLα-tag sequences and 92% for full-length PfEMP1 sequences, thereby significantly outperforming existing tools. Additionally, we developed a model to distinguish internal from subtelomeric var genes, which we applied to a global collection of P. falciparum genomes, revealing a higher frequency of internal var genes in Asia. upsML is available at https://github.com/sii-scRNA-Seq/upsML, providing a robust and efficient resource for large-scale var gene analysis. It can classify var genes from 20 genomes in under one second.
17 Apr 2026
Read appraisal →Proceedings of the National Academy of Sciences of the United States of America
Coalescence and translation: A language model for population genetics.
Probabilistic models such as the sequentially Markovian coalescent have long provided a powerful framework for population genetic inference, enabling reconstruction of demographic history and ancestral relationships from genomic data. However, these methods are inherently specialized, relying on predefined assumptions and/or limited scalability. Recent advances in simulation and deep learning provide an alternative approach: learning directly to generalize from synthetic genetic data to infer hidden evolutionary processes. Here we reframe the inference of coalescence times as a problem of translation between two biological languages: the sparse, observable patterns of mutation along the genome and the unobservable ancestral recombination graph that gave rise to them. Inspired by large language models, we develop cxt, a decoder-only transformer that autoregressively predicts coalescent events conditioned on local mutational context. We show that cxt performs competitively with state-of-the-art Markov Chain Monte Carlo-based likelihood models across a broad range of demographic scenarios, matching their accuracy in-distribution and approaching it in out-of-distribution, with the potential for improvement via fine-tuning. Trained on simulations spanning the stdpopsim catalog [J. R. Adrion et al., eLife 9, e54967 (2020); M. E. Lauterbur et al., eLife 12, e84874 (2023); G. Gower et al., Accessible, realistic genome simulation with selection using stdpopsim. bioRxiv [Preprint] (2025). https://doi.org/10.1101/2025.03.23.644823.], the model generalizes robustly and enables efficient inference at scale, producing over a million coalescence predictions in minutes. cxt also produces well-calibrated approximate posteriors, enabling principled uncertainty quantification. We apply cxt to population genomic data from both humans and mosquitoes, highlighting the model's ability to deal with the complexities of empirical data.
17 Apr 2026
Read appraisal →Physical chemistry chemical physics : PCCP
TransG4: an interpretable deep-learning approach for sequence-based G-quadruplex prediction
G-quadruplexes (G4) are non-canonical nucleic acid secondary structures formed in guanine-rich regions and have been shown to regulate diverse cellular processes such as gene expression, DNA replication, and telomere maintenance, with increasing evidence linking G4 to cancer and other human diseases. G4 predominantly emerge in guanine-rich regions and are implicated in a spectrum of molecular interactions and disease phenotypes, thus researchers are interested in the formation of G4. However, predicting the formation of G4 from nucleotide sequences is a persistent problem. Existing computational tools for G4 prediction are either rule-based on domain knowledge or rely on a single neural network model like a convolutional neural network (CNN), which lacks interpretability and struggles to capture long-range dependencies among bases. Here, we introduce TransG4, a novel neural network architecture that integrates a CNN, a transformer, and bidirectional gated recurrent units (BiGRUs) to identify potential G4 structures. TransG4 demonstrates strong predictive performance on both G4-seq and rG4-seq datasets, accurately predicting DNA mismatch scores and consistently outperforming existing methods in RNA RSR-ratio prediction. Attention-based interpretations further show that TransG4 captures biologically meaningful motifs consistent with canonical G4 structures, providing an interpretable and generalizable framework and representing a novel and impactful contribution to sequence-based G4 propensity prediction.
17 Apr 2026
Read appraisal →Computers in biology and medicine
A systematic review and benchmarking of modern metagenomic tools for taxonomic classification
Advancements in metagenomics have been driven by the continuous development of bioinformatic tools, particularly taxonomic classification software, which are central to the accurate characterization of microbial communities. However, establishing direct comparisons between these tools remains challenging due to variations in evaluation metrics, reference databases, and input data types. In this study, we present a systematic review of recently developed metagenomic taxonomic classification tools. Of the 31 identified tools, nine satisfied all functional and methodological criteria for the benchmark analysis. We evaluated their accuracy and computational performance using a standardized dataset derived from the NCBI RefSeq database. Our analysis revealed that most of these tools are domain-specific, each excelling in particular areas. Tools like TAMA, CAMITAX and PhyloFlash achieved higher accuracy for prokaryotic organisms, while ViWrap and PhaBOX achieved higher accuracy for viral classifications. SqueezeMeta achieved high F1 scores across most domains, though its assembly-based architecture limits effectiveness on highly diverse samples. MegaPath-Nano was least affected by increased mutation rates. The execution time varied widely among the tools, with domain-specific and machine learning-based tools generally being faster, while tools like BASTA had longer runtimes and lower accuracy. This review synthesizes performance results for current tools, providing an overview of their strengths and computational methodologies.
17 Apr 2026
Read appraisal →PloS one
Exploration of schizophrenia-associated gene modules using graph theory, co-expression networks, and dimensionality reduction.
The PsychENCODE consortium has generated a comprehensive multi-omic dataset from human brain samples, spanning healthy individuals and those with neuropsychiatric disorders. In this study, we focus exclusively on schizophrenia and analyze PsychENCODE transcriptomic data to construct gene regulatory and co-expression networks, aiming to uncover biologically and clinically relevant gene modules. We apply three independent analytical approaches to the same dataset. First, we preprocess the data using Surrogate Variable Analysis (SVA) to correct for latent variation and normalize gene expression. For graph-based analysis, we use Pearson correlation and the igraph package to construct gene co-expression networks and apply Prim's algorithm to generate minimum spanning trees (MST) and compute centrality measures. Next, we implement Weighted Gene Co-expression Network Analysis (WGCNA), assuming a scale-free topology, to identify modules associated with schizophrenia traits. Dimensionality reduction is performed using Principal Component Analysis (PCA), with visualization aided by t-distributed Stochastic Neighbor Embedding (t-SNE) and Multidimensional Scaling (MDS). Functional enrichment is carried out using Gene Ontology (GO) and KEGG pathway databases. Each method reveals distinct yet complementary biological signatures associated with schizophrenia. The igraph-based approach highlights differentially expressed genes (DEGs) with high centrality, uncovering hub genes involved in mRNA export, synaptic signaling, and ion transport. WGCNA identifies co-expression modules strongly correlated with schizophrenia diagnosis, enriched for immune response, histone modification, and mRNA surveillance pathways. PCA isolates key genes contributing to diagnostic variance, with enrichment in neurotransmitter release cycles and cytokine signaling. Collectively, these results underscore the involvement of immune, synaptic, and epigenetic processes in schizophrenia and demonstrate the power of using multiple, orthogonal analytical lenses.
17 Apr 2026
Read appraisal →Computers in biology and medicine
Deep learning with limited data: a transfer learning approach for transcriptomic survival prediction
BACKGROUND: Deep neural networks (DNNs) are promising for analyzing high-dimensional transcriptomic data in cancer research but are limited by data scarcity and heterogeneity. Transfer learning (TL), which leverages large datasets to improve performance on smaller ones, has been widely applied in classification and clustering, yet its use in survival prediction remains underexplored. METHODS: We evaluated transfer learning for disease-free survival prediction using RNA-seq data from 7509 patients across 27 tumor types from The Cancer Genome Atlas. A pan-cancer DNN was optimized via random search across 750 configurations. Tumor-specific baseline models were trained independently, while transfer learning models were built with a leave-one-out pre-training strategy followed by fine-tuning on the target tumor. Performance was assessed with the concordance index. Biological relevance was evaluated through a LIME-like interpretability approach and gene-set enrichment analysis. RESULTS: Transfer learning outperformed tumor-specific baselines in 24 of 27 tumor types, with the greatest improvements in smaller cohorts. Fine-tuning consistently enhanced predictions by adapting pan-cancer features to tumor-specific data. Enrichment analysis confirmed that the models captured both general oncogenic mechanisms and tumor-specific pathways. CONCLUSIONS: Transfer learning improves survival prediction from transcriptomic data, especially in small cohorts, by exploiting molecular patterns shared across cancers. These findings support transfer learning as a robust strategy for survival modeling and a step toward foundation models in molecular oncology.
16 Apr 2026
Read appraisal →International immunopharmacology
CXCL9 as a key feature for deep learning-based immune subtyping and prediction of immune checkpoint blockade response in triple-negative breast cancer
BACKGROUND: Immune checkpoint blockade (ICB) therapy has provided a promising treatment option for patients with triple-negative breast cancer (TNBC). However, existing efficacy prediction strategies based on programmed death-ligand 1 (PD-L1) expression and Tumor Mutational Burden (TMB) have significant limitations, leading to inaccurate patient selection. METHODS: This study integrated multi-omics data obtained from GEO, TCGA, and GTEx datasets. Based on ComBat correction results, two deep learning-driven unsupervised clustering methods were constructed to identify potential immune subtypes. These identified TNBC subtypes were subjected to comprehensive bioinformatics analyses to investigate their differences in overall survival, immune cell infiltration, and functional enrichment analysis, which clarified the biological rationale of the proposed classification. Subsequently, differential expression analysis (DEA) and multiple machine learning algorithms were applied to identify subtype-specific features, with the expression and differentiation status of the key feature in immune cells further validated using single-cell sequencing (scRNA-seq). Finally, co-expression networks at multiple cellular levels were constructed to predict the co-expressed regulatory targets of the key feature, and the hypothesized regulatory mechanism was preliminarily validated in in vitro experiments (qRT-PCR, western blotting, immunofluorescence, and ELISA). RESULTS: After 5-fold cross-validation, the AE-K-means clustering method classified TNBC into three distinct latent subtypes. This algorithm exhibited superior classification efficacy compared with the other models (including NMF, ConsensusClusterPlus and VAE-GMM), as demonstrated by markedly elevated Silhouette Coefficients (training: 0.585 ± 0.030; test: 0.698 ± 0.103; validation: 0.611) and substantially reduced Davies-Bouldin Index values (training: 0.623 ± 0.032; test: 0.385 ± 0.218; validation: 0.558) at K = 3. These three identified TNBC subtypes displayed significant differences in overall survival and exhibited pronounced immune heterogeneity. Specifically, the K2 subtype was characterized by high PD-L1 expression and abundant M1 macrophage infiltration, while the K3 subtype displayed an opposing, immunosuppressive microenvironment. Feature selection results from multiple machine learning models were intersected to identify key features including CXCL9, and an RF model constructed based on these intersected genes showed high efficacy, particularly in distinguishing the K2 and K3 subtypes (AUC = 0.941). Further scRNA-seq analysis revealed that CXCL9 was specifically highly expressed in myeloid cells, and the proportion of CXCL9+ macrophages differed significantly between responders and non-responders. Given the strong positive correlation between CXCL9 and IDO1 expression, we subsequently treated macrophages with IDO1 inhibitors and observed a marked upregulation in both the expression and secretion of CXCL9, suggesting that IDO1 overexpression in macrophages may represent a potential mechanism underlying CXCL9 exhaustion in the tumor microenvironment. CONCLUSIONS: Our study developed a novel immune classification system for TNBC, identifying CXCL9 as a key biomarker positively correlated with ICB response and regulated by IDO1. These findings provide a basis for precision immunotherapy in TNBC and support the exploration of IDO1 inhibitors in combination with ICB.
15 Apr 2026
Read appraisal →Journal of molecular modeling
HopWD-DTA: a novel framework for drug-target affinity prediction fusing multi-hop neighborhoods and deep features
CONTEXT: Accurate prediction of drug-target affinity (DTA) is crucial for accelerating drug discovery, but it remains a significant challenge. While deep learning methods have shown promise, many existing models struggle with representing long-range protein relationships within protein-protein interaction (PPI) networks, often suffering from performance degradation as graph neural network layers deepen. To overcome this limitation, we developed HopWD-DTA, a novel framework designed to effectively capture both local and global protein features. The core innovation lies in integrating multi-hop neighborhood information from PPI networks with deep structural features of proteins and drugs. Comprehensive evaluations on the Davis, KIBA, and Human benchmark datasets demonstrate that HopWD-DTA achieves state-of-the-art performance, significantly improving DTA prediction accuracy over existing cutting-edge solutions. METHODS: The HopWD-DTA framework consists of separate branches for protein and drug feature extraction. For proteins, structural features from residue contact maps are first encoded using a Graph Convolutional Network (GCN). These features, along with InterPro annotations, are then embedded into a PPI network. We introduce a multi-hop neighborhood serialization technique, which generates a sequence of feature matrices representing different neighborhood scopes. This sequence is processed by a Variational Autoencoder (VAE) to learn a robust protein representation. For drugs, molecules are represented as graphs from SMILES strings and encoded via a GCN, followed by a novel Wide-and-Deep Path (WDPATH) module to capture both macroscopic and microscopic features. The final protein and drug features are concatenated and fed into a multilayer perceptron for affinity prediction. The model was implemented using PyTorch and RDKit.
15 Apr 2026
Read appraisal →Journal of chemical information and modeling
zERExtractor: An Automated Platform for Enzyme-Catalyzed Reaction Data Extraction from Scientific Literature
The rapid expansion of enzyme reaction literature has created a major bottleneck in database curation, leaving vast amounts of enzyme-substrate-condition relationships unstructured and inaccessible for DL-driven modeling. How to fully utilize the enzymatic reaction data has been an important task for future accurate enzyme activity prediction models. Current deep learning (DL)-based data extraction models heavily rely on large language models (LLMs) without a fidelity check and the ability to continuously evolve. To address these issues, we developed zERExtractor (Zelixir's Enzyme Reaction Data Extractor), an accuracy-oriented and extensible platform for extracting enzyme-catalyzed reaction data from scientific publications. This system offers a unified multimodal information extraction framework (covering molecular reaction diagrams, tables, and texts) to integrate enzymatic reaction descriptors into structured storage. We employ fine-tuned large LLMs together with DL in a human-in-the-loop pipeline that evolves through data fidelity validation by experts and active learning. Also, zERExtractor achieves 89.9% accuracy in table recognition and over 98% accuracy in molecular image recognition on synthetic data sets, outperforming the strongest baseline by more than 2% and consistently maintaining above 95% on realistic benchmarks. zERExtractor bridges the data gap in enzyme reaction data with a scalable framework for accurate multimodal extraction, advancing DL-driven enzyme modeling and enabling future applications in computational enzymology and biotechnology. The platform is publicly accessible online at https://zpaper.zelixir.com/.
14 Apr 2026
Read appraisal →Journal of chemical information and modeling
AI-MedCraft: A Strategy-Driven AI Platform for Multi-Objective Molecular Design
Small-molecule drug discovery requires balancing potency, selectivity, solubility, safety, and synthetic feasibility, among other interdependent properties. Yet most computational workflows address these properties sequentially or collapse them into fixed weighted scores. Although artificial intelligence has accelerated molecular design, many platforms struggle to handle explicit multiobjective trade-offs. Here, we present AI-MedCraft, a strategy-driven molecular design framework that applies adaptive, Pareto-guided reinforcement learning to optimize multiple objectives concurrently within a unified workflow. When structural information is available, physics-aware scoring is incorporated to support binding-competent designs. In a structure-based benchmark on the solubility rescue of the BTK inhibitor GDC-0834, AI-MedCraft achieves broader Pareto-front coverage than the scalar-reward reinforcement learning framework REINVENT 4 under matched objectives and computational cost. In a second case study, AI-MedCraft redesigns Efavirenz to retain HIV-1 reverse transcriptase engagement while reducing predicted 5-HT2A off-target binding, demonstrating its use in multiconstraint molecular optimization.
14 Apr 2026
Read appraisal →Journal of chemical information and modeling
ncProFormer: A CNN-enhanced Transformer for ncRNA Coding-Potential Prediction
Recent studies have revealed that some noncoding RNAs (ncRNAs) bear translational potential, and their encoded micropeptides have essential functions in multiple biological processes. However, accurate identification of coding-capable ncRNAs remains challenging due to weak translation signals, low conservation, and heterogeneous data distributions. Herein, we propose ncProFormer, a deep learning framework tailored for ncRNA coding-potential prediction. ncProFormer integrates the nucleic-acid language model GENA-LM to obtain contextual sequence embeddings, adopts an all-token representation strategy, and employs a convolutional neural network (CNN)-enhanced transformer encoder to jointly capture local nucleotide patterns and long-range dependencies. ncProFormer consistently outperformed the existing methods across the in-house human data set, the external validation data set, the public CPPred benchmark data set. More importantly, this study presents the first cross-species evaluation in ncRNA coding-potential prediction. Without retraining, ncProFormer maintained its strong predictive performance on mouse and rat data sets, showing that the learned biological representations are transferable and it is robust under the distributional shift and cross-species conditions. Collectively, these findings establish ncProFormer as an effective and generalizable framework for uncovering the coding potential of ncRNAs, thus offering a promising computational tool for characterizing ncRNA functions across diverse transcriptomic contexts.
14 Apr 2026
Read appraisal →Journal of chemical information and modeling
ToxPLTC: Peptide Toxicity Prediction by Integrating Pretrained T5 Protein Language Model and Text Convolutional Neural Network
Peptide-based therapeutics show promising potential in treating a range of diseases, such as diabetes, cancer, and chronic pain. However, critical challenges, including peptide toxicity, immunogenicity, and stability deficiencies of peptides, have become major obstacles to their direct clinical application. Traditional toxicity testing methods based on a wet lab are not only costly but also time-consuming. In contrast, the classification model based on deep learning provides a new technical path for the efficient identification of peptide toxicity. In this study, we propose a deep learning framework called ToxPLTC, which employs the ProtT5 protein language model based on the Transformer architecture for pretraining peptide sequences, adopts the borderline SMOTE algorithm to handle an imbalanced training set data, and utilizes a text convolutional neural network combined with a fully connected layer for classification. Additionally, visualization analysis, motif analysis, and mutation-scan analysis are performed to understand the function of each module and enhance the interpretability of our model. The applicability domain is constructed based on the K-NN strategy to define the effective prediction range of our model to ensure the reliability of model predictions. The ToxPLTC model achieves a balanced accuracy of 93.02% on independent test set 1 and 88.04% on independent test set 2. Experimental results demonstrate that our model outperforms existing models on independent test sets and has good generalization ability. The ToxPLTC model possesses great potential as a valuable and robust tool for peptide-based drug development. The source data sets and codes can be available at the following GitHub repository: https://github.com/yunyunliang88/ToxPLTC.
14 Apr 2026
Read appraisal →Medicine
Glioma biomarker analysis and drug targeting prediction based on ferroptosis-related genes
Glioma represents a significant challenge in neuro-oncology due to its highly invasive nature and poor prognosis. Recent studies indicate that ferroptosis, a form of regulated cell death, plays a crucial role in the pathogenesis of glioma. Modulating ferroptosis may offer a novel therapeutic approach for glioma treatment. This study aims to identify ferroptosis-related biomarkers and patterns of immune infiltration in glioma using bioinformatics techniques, and to predict potential therapeutic drugs targeting these biomarkers. Gene expression profiles (GSE16011, GSE50161) were retrieved from the GEO database, while ferroptosis-related genes were sourced from the FerrDb database. Intersection genes were identified using the CytoHubba algorithm and a consensus clustering method. Immune infiltration was assessed using the CIBERSORT algorithm. Functional enrichment analysis and gene set enrichment analysis were conducted with Gene Ontology and Kyoto Encyclopedia of Genes and Genomes databases. Drug targeting predictions were made based on identified hub genes. We identified 10 hub genes: TP53, RRM2, EZH2, CDKN1A, MYCN, KIF20A, BLM, GLS2, HMOX1, and GOT1. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes analyses revealed that ferroptosis is primarily associated with apoptosis, reactive oxygen metabolism, the p53 signaling pathway, and immune regulation. Immune infiltration analysis showed significant differences in the expression of plasma cells, CD8 T cells, follicular helper T cells, activated NK cells, and macrophages between groups. qRT-PCR validation in glioma cell lines confirmed the differential expression of key ferroptosis-related genes, supporting the robustness of our bioinformatic findings. Furthermore, molecular docking demonstrates excellent binding properties between the drug components and key targets, and these key targets were further validated by PCR analysis in vitro. In summary, this study highlights the potential of ferroptosis-related biomarkers in glioma and suggests novel drug targets, providing a foundation for future therapeutic strategies.
12 Apr 2026
Read appraisal →PloS one
Integrative bioinformatics and machine learning identify shared molecular mechanisms and diagnostic biomarkers between Helicobacter pylori infection and atrial fibrillation
BACKGROUND: Helicobacter pylori (H. pylori) infection and atrial fibrillation(AF) are major global health concerns. Emerging evidence has suggested a potentially chronic inflammation-mediated link between them, but the shared genetic mechanisms remain unclear. METHODS: We analyzed multidataset gene expression profiles from the Gene Expression Omnibus (GEO) database. Differential expression analysis, weighted gene co-expression network analysis (WGCNA), functional enrichment, and machine learning were employed to identify common genes, pathways, and diagnostic biomarkers. Protein-protein interaction (PPI) networks, drug-gene analysis, and molecular docking were used to identify hub genes and potential therapeutics. RESULTS: We identified 73 common differentially expressed genes (DEGs) between H. pylori infection and AF, which were predominantly enriched in immune-related processes including leukocyte activation, neutrophil migration, and myeloid cell-mediated immunity. Machine learning identified 15 and 23 key feature genes for H. pylori and AF, respectively, with S100A8 emerging as a shared diagnostic biomarker. Ten hub genes including TYROBP, ITGB2, and SPI1, were identified from the PPI network. Drug repositioning analysis suggested retinoic acid, indirubin, and ropivacaine as candidate therapeutics targeting these key hub genes. CONCLUSION: Our integrative analysis highlights the central role of immune-inflammatory pathways in linking H. pylori infection to AF. We propose S100A8 and other identified hub genes as potential biomarkers and therapeutic targets. The predicted candidate therapeutics, particularly retinoic acid, may offer novel avenues for intervention, warranting further experimental validation.
12 Apr 2026
Read appraisal →HGG advances
Gene-specific pathogenicity predictor for chromatin remodeling BAF complex-associated neurodevelopmental disorders
Advancements in whole-genome sequencing have increased the number of variants of uncertain significance (VUS) identified in human genomes. This has created a diagnostic bottleneck for genetic counselors tasked with sifting through these variants and determining those most likely to be causative for a patient's clinical presentation. Machine learning (ML) tools can aid in identifying pathogenic variants from VUS, but there is a need for gene-specific algorithms that predict pathogenic variants with high accuracy. To address this need, we present a workflow for developing gene-specific, ensemble-learning ML tools, that leverage outputs from other algorithms, locations of variants within the gene, and evolutionary conservation data to make a prediction of pathogenicity. Variants in SMARCA2 and SMARCA4 that are associated with rare neurodevelopmental diseases were used to screen 15 ML algorithms. A random forest learner was tuned to yield a final accuracy of 0.93 on holdout data. Generalizing this predictor to other BRG1/BRM-associated factor (BAF) complex proteins resulted in a sharp decline in performance. We trained a final predictor for all genes in the study to create a predictor that identifies pathogenic variants in these BAF subunits with an accuracy of 0.91 on holdout data. This predictor specific to BAF complex proteins performs with higher accuracy and area under the precision-recall curve than any other predictor. The decline in performance when generalized to other proteins emphasizes the need for the gene-specific calibration of predictors. Our workflow for the development of such models provides a quick, computationally inexpensive route for improving the ML tools available to genetic counselors.
12 Apr 2026
Read appraisal →PloS one
Predicted meta-omics: A potential solution to multi-omics data scarcity in microbiome studies
Imbalances in the gut microbiome have been linked to conditions such as inflammatory bowel disease, diabetes, and cancer. While metagenomics and amplicon sequencing are commonly used to study the microbiome, they do not capture all layers of microbial functions. Other meta-omics data can provide more insights, but these are more costly and laborious to procure. The growing availability of paired meta-omics data offers an opportunity to develop machine learning models that can infer connections between metagenomics data and other forms of meta-omics data, enabling the prediction of these other forms of meta-omics data from metagenomics. We evaluated several machine learning models for predicting meta-omics features from various meta-omics inputs. Simpler architectures such as elastic net regression and random forests generated reliable predictions of transcript and metabolite abundances, with correlations of up to 0.77 and 0.74, respectively, but predicting protein profiles was more challenging. We also identified a core set of well-predicted features for each meta-omics output type, and showed that multi-output regression neural networks performed similarly when trained using fewer output features. Lastly, our experiments demonstrated that predicted features can be used for the downstream task of inflammatory bowel disease classification, with performance comparable to that of experimental data.
11 Apr 2026
Read appraisal →PLoS medicine
Spatial omics and AI for clinically actionable cancer biomarkers
Integrating spatial omics with artificial intelligence is likely to advance biomarker research and diagnostics, with the potential to pair mechanistic insight into spatial target biology. By developing scalable, reproducible quantification in routine pathology, it can help bridge discovery, validation, and real-world clinical implementation.
11 Apr 2026
Read appraisal →Cell genomics
Variant-resolved prediction of context-specific isoform variation with a graph-based attention model
In eukaryotes, most genes produce multiple transcript isoforms that diversify the transcriptome and proteome, serving as a key mechanism of functional regulation. Genetic variation can disrupt the RNA processing signals that shape isoform structure and abundance, yet modeling these effects at full-length isoform resolution remains challenging due to the complexity of transcript regulation. Here, we introduce Otari, an attention-based graph neural network framework trained on the human genomic sequence and long-read transcriptomes across 30 tissue types and brain regions. Otari predicts tissue-specific differential isoform abundance by integrating sequence-derived epigenetic and post-transcriptional signals, enabling isoform-resolved variant effect interpretation. Applied to large-scale variant datasets, including an autism cohort, Otari uncovers patterns of isoform dysregulation undetectable at the gene level, such as variant-driven perturbations in isoform abundance and microexon usage implicated in autism pathophysiology. We provide Otari as a resource for powering isoform-level analyses across tissues at scale.
11 Apr 2026
Read appraisal →Microbiology spectrum
PneumoWiki: a pan-genome-based database for the pathogen Streptococcus pneumoniae
UNLABELLED: The Gram-positive bacterium Streptococcus pneumoniae is a major human pathogen that exhibits more than 100 different capsular serotypes and considerable genomic variability. S. pneumoniae is also an important model organism for basic and biomedical research, with a number of strains with different characteristics being used. To address this, we have created the manually curated pan-genome database PneumoWiki (https://pneumowiki.med.uni-greifswald.de) that integrates genomic data of 43 S. pneumoniae strains with various aspects of functional annotation. All data relating to a specific gene and gene product are compiled on a Gene page. In addition to the gene and protein sequences and annotation from NCBI RefSeq, this includes, for example, data on gene expression, protein function, and localization. The information is supplemented with links to the data sources as well as to other databases and relevant literature. The basic concept of PneumoWiki is the interlinked presentation of pan-genome-based and strain-specific information. Therefore, register tabs with strain names at the top of the Gene pages allow switching between orthologous genes and the corresponding pan-genome page. The pan-genome gene pages contain a summary with functional information and the occurrence of the gene, an ortholog table for the 43 S. pneumoniae strains, a multiple-strain genome viewer as well as a protein sequence alignment. PneumoWiki was developed as a user-friendly tool and is available free of charge. The data collected in PneumoWiki are accessible via various download options. Overall, PneumoWiki is a resource for the research community that supports the analysis and interpretation of experimental data and, in particular, enables the integration of the knowledge available for different S. pneumoniae strains. IMPORTANCE: Databases that integrate and curate the large amounts of genomic data and functional information available for well-studied organisms play an increasingly important role in many areas of life sciences, including infection research. Streptococcus pneumoniae is a major human pathogen associated with high morbidity and mortality rates. The species exhibits high variability in its genome sequence, and a number of strains with different characteristics are used in pneumococcal research. We created the wiki-type database PneumoWiki to provide researchers with curated, up-to-date information on a range of pneumococcal genomes, focusing on the individual genes and encoded proteins and their functional annotation. The key feature of PneumoWiki is the pan-genome approach, which allows combining information of individual strains, thereby supporting the analysis of experimental data and comparisons between studies using different strains. In this way, the database becomes a valuable research tool that can contribute to a better understanding of pneumococcal physiology.
10 Apr 2026
Read appraisal →Chemical communications (Cambridge, England)
Innovative integration of molecular docking and machine learning for drug discovery: from virtual screening to nanomolar inhibitors
This review highlights our group's systematic approach to integrating molecular docking, pharmacophore modeling, and machine learning methodologies for the rational discovery of bioactive leads. We describe innovative strategies including docking-based data augmentation, ligand-receptor contact fingerprints, genetic algorithm-guided feature selection, and SHAP-based model interpretation. These approaches have enabled the discovery of nanomolar inhibitors against multiple therapeutic targets including STAT3, TTK, LSD-1, and HER2. The presented workflow demonstrates how machine learning (ML) can be synergistically combined with traditional computer-aided drug design methods to achieve efficient scaffold hopping and identify novel chemotypes with potent biological activities.
10 Apr 2026
Read appraisal →Analytical chemistry
SIBioX: A Matrix Based Bioinformatics Analysis Tool Based on Swarm Intelligence Algorithm
Biological matrix data are essential for computational analysis, providing a structured framework to identify patterns and relationships in biological systems. Many other biological data types, including sequences, networks, and images, can be transformed into matrix representations through feature extraction and encoding. However, their high dimensionality complicates analysis, leading to increased computational complexity and the risk of overfitting, known as the curse of dimensionality. To address these challenges, we developed SIBioX, a matrix-based bioinformatics tool powered by swarm intelligence algorithms. It integrates 54 swarm intelligence methods, 5 conventional feature selection techniques, and 17 machine learning models, enabling comprehensive analysis of biological matrix data. With a user-friendly graphical interface, it supports operations such as feature normalization, selection, classification, clustering, statistical analysis, and data visualization. Additionally, it converts nonmatrix biological data, like gene and protein sequences, into matrix formats for further study. Experimental results demonstrate that SIBioX not only attains high accuracy in feature selection but also effectively reduces dimensionality, thereby streamlining bioinformatics workflows and promoting greater efficiency in biomedical research.
10 Apr 2026
Read appraisal →Genome research
spRefine denoises and imputes spatial transcriptomic data with a reference-free framework powered by genomic language model
The analysis of spatial transcriptomic data is hindered by high noise levels and missing gene measurements, challenges that are further compounded by the higher cost of spatial data compared to traditional single-cell data. To overcome this challenge, we introduce spRefine, a deep learning framework that leverages genomic language models to jointly denoise and impute spatial transcriptomic data. Our results demonstrate that spRefine yields more robust cell- and spot-level representations after denoising and imputation, substantially improving data integration. In addition, spRefine serves as a strong framework for model pretraining and the discovery of novel biological signals, as highlighted by multiple downstream applications across data sets of varying scales. Notably, spRefine enhances the accuracy of spatial aging clock estimations and uncovers new aging-related relationships associated with key biological processes, such as neuronal function loss, which offers new insights for analyzing aging effect with spatial transcriptomics.
8 Apr 2026
Read appraisal →Proceedings of the National Academy of Sciences of the United States of America
Proteome-wide prediction of interactions between structured domains and peptide motifs reveals functionally coherent subnetworks
Many protein-protein interactions (PPIs) are mediated by the binding of short linear motifs (SLiMs) to peptide recognition domains (PRDs). Here, we describe PrePPI-SLiM, a proteome-scale computational pipeline that leverages data from the Eukaryotic Linear Motif (ELM) database to predict whether two proteins will form a peptide-mediated complex. The ELM database defines classes of protein-peptide interactions with SLiMs represented by sequence motifs and PRDs represented by Pfam domains. PrePPI-SLiM systematically evaluates all pairwise combinations of proteins within a proteome and identifies PRD-SLiM pairs that occur in the same ELM class. This evidence together with disorder prediction and sequence conservation of the motif are integrated in a naïve Bayes framework to assign a likelihood for complex formation. To obtain potential PDB templates for atomistic models of PrePPI-SLiM interactions, we associate individual PPI predictions with homologous PDB complexes involving the same PRD Pfam domain and SLIM, and obtain PDB templates for 92% of our high-confidence predictions. Moreover, studies with AF3Complex suggest that prior knowledge of the interacting PRD and SLiM, as provided here, is a critical starting point for creating a 3D model of the specific sequences of the PRD and SLiM query proteins. Finally, we demonstrate that clustering of the high-confidence PrePPI-SLiM interactome yields functionally coherent PPI networks that reveal mechanistic insights into cellular processes. The PrePPI webserver provides convenient access to high-confidence PrePPI-SLiM predictions, PDB templates for modeling, and functional networks.
8 Apr 2026
Read appraisal →Neuroscience
Limitations and opportunities in multi-omics integration for neurodevelopmental, neurodegenerative and psychiatric disorders: A systematic review
Recent advances in high-throughput technologies have led to an increased generation of biological data across genomics, transcriptomics, proteomics, epigenomics, and metabolomics. However, a major challenge remains: effectively integrating these multi-omics datasets to allow a more holistic understanding of the complex, interconnected mechanisms underlying human diseases. Neurodevelopmental, neurodegenerative, and psychiatric disorders are particularly multifactorial and heterogeneous, making them candidates for multi-omics approaches. In this context, this systematic review assesses the current state of multi-omics integration in neurological research. Records retrieved from five major databases were processed, and 156 studies were included for further analysis. The most frequently studied conditions were Alzheimer's Disease, Depressive Disorder and Parkinson's Disease, with epigenomics-transcriptomics and metagenomics-metabolomics emerging as the most common omics pairings. The field remains dominated by studies integrating pairs of omics layers. Only a limited number of computational tools are currently being applied to the integration of more than two omics layers, highlighting a gap in comprehensive multi-omics modeling. Despite progress, key challenges persist, including data accessibility and the need for standardized frameworks to allow cross-study comparisons. Moreover, most computational findings lack experimental validation in wet-laboratory settings. Future research should address these challenges, develop scalable algorithms for integrating multi-omics data, and leverage large, open-access datasets. Integrating computational predictions with experimental validation could help researchers prioritize high-confidence biomarkers relevant to clinical applications. Collaborative efforts among bioinformaticians, clinicians, and experimentalists will be essential to translating these advances into clinically actionable solutions.
6 Apr 2026
Read appraisal →Clinical genetics
Clinical Feasibility of Long-Read WGS for DNA Methylation Signature Analysis
DNA methylation (DNAm) signatures have emerged as valuable diagnostic biomarkers for rare genetic disorders. To date, the most widely used approach for establishing and validating these signatures has relied on array-based technologies. However, in clinical diagnostics, there is a growing shift from short-read sequencing (SRS) toward long-read sequencing (LRS) technologies. Recent advances in platforms such as Pacific Biosciences (PacBio) and Oxford Nanopore Technologies (ONT) enable direct assessment of DNAm from native DNA, offering improved resolution and reduced technical bias compared to array-based technologies. In this study, we compared DNAm profiles generated by LRS with those obtained from DNAm arrays. DNAm profiles of two individuals with pathogenic KMT2D variants were analyzed using DNAm arrays, LRS using PacBio and ONT, and ONT multiplexed sequencing with adaptive sampling. A support vector machine (SVM) classifier trained on array data, as well as the public classification platform EpigenCentral, yielded correct predictions for all LRS samples, underscoring the potential of LRS platforms in DNAm-based diagnostics. Our results suggest that DNAm profiles generated by LRS align well with DNAm signatures established using DNAm arrays, supporting their feasibility in clinical and research applications with the added benefit of simultaneous methylation and sequence analysis.
4 Apr 2026
Read appraisal →Critical reviews in oncology/hematology
Artificial intelligence and multi-omics convergence in breast cancer: Revolutionizing diagnosis, prognostication, and precision oncology
Breast cancer (BC) is a highly heterogeneous malignancy and remains a major cause of cancer-related mortality among women worldwide. Advances in multi-omics profiling spanning genomics, transcriptomics, epigenomics, proteomics, and metabolomics have enabled finer subtype stratification and more comprehensive characterisation of tumour biology, thereby accelerating the discovery of diagnostic and prognostic biomarkers and actionable therapeutic targets. Nonetheless, translating multi-layer molecular signals into clinically robust decision support remains challenging because of the high dimensionality and heterogeneity of omics data, cross-cohort and cross-platform variability, and the fragmentation inherent to single-modality analyses. This review summarises how multi-omics studies have refined BC subtype definitions and advanced biomarker and target identification, and then synthesises recent progress in artificial intelligence, particularly deep learning, for integrating multi-omics with imaging, pathology, and clinical variables to improve diagnosis, risk stratification, prognosis prediction, and treatment response assessment. We critically examine representative multimodal integration frameworks and emerging deep learning architectures that learn both shared and modality-specific representations, which in many settings enable more accurate patient-level prediction than unimodal baselines. We further delineate key barriers to clinical translation, including cross-centre heterogeneity and inconsistent endpoint definitions, structural missingness of modalities in real-world workflows, inadequate cross-platform normalisation, limited interpretability and auditability, and a lack of prospective validation. Finally, we propose realistic next steps, including standardised and auditable preprocessing pipelines, missingness-aware fusion strategies, explainable and uncertainty-aware modelling, privacy-preserving multi-centre learning, and prospective, workflow-based evaluation. Collectively, these perspectives provide a roadmap for advancing multimodal AI-multi-omics integration toward reliable clinical deployment in BC management.
4 Apr 2026
Read appraisal →The Journal of investigative dermatology
A Metabolism-Driven Prognostic Model and PSMD14-SP1-GYS1 Axis Reveal Therapeutic Vulnerabilities in Melanoma
Melanoma is a highly aggressive cutaneous malignancy characterized by a strong propensity for metastasis and therapy resistance, with its progression being closely linked to metabolic reprogramming. This study integrated multiomics data (The Cancer Genome Atlas, Gene Expression Omnibus, European Nucleotide Archive) and advanced machine learning to develop prognostic and immunotherapy prediction models for melanoma, focusing on 114 metabolism-related pathways. Cox regression identified 70 genes linked to survival, with functional enrichment revealing key metabolic pathway alterations. A metabolism-related prognostic model (MRPM) was constructed using 101 combinations of machine learning algorithms, demonstrating superior predictive accuracy across 4 cohorts. High-risk patients showed worse survival and immunotherapy response in melanoma and other cancers. Tumor microenvironment analysis revealed MRPM's negative correlation with immune infiltration and positive association with tumor purity. Single-cell sequencing highlighted MRPM gene enrichment in melanocytes. Mechanistically, GYS1 (the key gene in MRPM) emerged as a pivotal prognostic gene that promotes melanoma proliferation and metastasis. Regulatory studies uncovered SP1's transcriptional control of GYS1- and PSMD14-mediated stabilization of SP1 through K48-linked ubiquitination removal. In vivo validation confirmed that PSMD14 knockdown suppressed tumor growth through SP1-GYS1 axis disruption. This work establishes MRPM as a robust predictive tool and elucidates the PSMD14-SP1-GYS1 regulatory network as a potential therapeutic target in melanoma metabolism.
4 Apr 2026
Read appraisal →Current opinion in genetics & development
Deep learning for psychiatric genomics: from tools to applications
The genetic architecture of psychiatric disorders is highly complex, with genome-wide association studies implicating thousands of risk loci. A central challenge is that most of these variants are located in noncoding regions, making it difficult to elucidate their regulatory consequences within the brain's intricate cellular landscape. The recent convergence of advanced artificial intelligence, particularly deep learning (DL), has catalyzed a paradigm shift by providing powerful tools to address this gap. This review traces the evolution of DL in genomics, beginning with task-specific models. We then examine the transformative impact of foundation models, pretrained neural networks that learn the 'language' of biology, including genomic language models, single-cell foundation models, and large language models originally trained on natural language. Finally, we survey applications to key problems in psychiatric genomics. We hope this review provides a comprehensive overview of recent advances in DL for genomics and serves as a bridge to help researchers in psychiatric genomics more effectively understand and apply these frontier methods to guide the development of novel therapeutic strategies for psychiatric disorders.
4 Apr 2026
Read appraisal →Int J Surg
Multi-omics dissection of high TWAS-active endothelial pathogenesis in pulmonary arterial hypertension: bridging single-cell heterogeneity, machine learning-driven biomarkers, and developmental reprogramming
BACKGROUND: Pulmonary Arterial Hypertension (PAH) is a leading cause of cardiovascular-related mortality worldwide. The emergence of single-cell RNA sequencing (scRNA-seq) has enhanced the ability to dissect cellular heterogeneity in PAH at a granular level. Transcriptome-wide association studies (TWAS) leverage expression quantitative trait loci (eQTL) and genome-wide association study (GWAS) data to identify novel susceptibility genes whose genetically predicted expression correlates with disease risk. However, no study has systematically integrated TWAS with scRNA-seq to unravel the pathogenesis of PAH at single-cell resolution. METHODS: Using TWAS analysis, we identified a set of candidate genes genetically associated with PAH. We then evaluated the differential activity of these genes across PAH cell types at single-cell resolution using AUCell, Ucell, ssGSEA (Single Sample Gene Set Enrichment Analysis), and AddModuleScore algorithms. A subset of endothelial cells exhibiting elevated TWAS activity was identified via quartile-based stratification and designated as the high TWAS activity state (HTS) group. Multi-dimensional analyses, including observed-to-expected ratio (RO/E), CellChat, CytoTRACE (CytoTRACE is based on the robust observation that transcriptional diversity decreases during cell differentiation), and scMetabolism, were employed to characterize the functional and communicative properties of HTS cells. Machine learning algorithms were integrated to identify signature genes of the HTS subpopulation, and a benchmarked random forest model was trained to predict HTS status. We performed immunohistochemistry and quantitative reverse transcription- polymerase chain reaction (qRT-PCR) validation of the signature genes (KLF2, RASIP1 and DEPP1) in PAH and control lung tissues to support their expression patterns. RESULTS: We demonstrated that HTS endothelial cells are strongly associated with PAH pathogenesis, exhibiting significant tissue tropism, enhanced roles in intercellular communication, and a progenitor-like function in endothelial differentiation. Machine learning-based feature selection revealed three robust signature genes: KLF2, RASIP1, and DEPP1. These genes demonstrated exceptional predictive power for identifying HTS cells, suggesting their potential as drivers of endothelial dysfunction in PAH. The random forest model, benchmarked against multiple algorithms, achieved high accuracy in predicting PAH progression using these genes. Immunohistochemical analysis of pulmonary artery and qRT-PCR result of lung tissues addressed the elevated expression of KLF2, RASIP1 and DEPP1 in arterial wall post-PAH. CONCLUSION: This study elucidates endothelial cell heterogeneity in PAH and establishes the central role of HTS cells in disease progression, cellular crosstalk, and developmental reprogramming. Our findings bridge the gap between GWAS and scRNA-seq methodologies and provide a transformative framework for understanding PAH mechanisms.
1 Feb 2026
Read appraisal →Artif Cells Nanomed Biotechnol
Unveiling the ageing-related genes in diagnosing osteoarthritis with metabolic syndrome by integrated bioinformatics analysis and machine learning
Ageing significantly contributes to osteoarthritis (OA) and metabolic syndrome (MetS) pathogenesis, yet the underlying mechanisms remain unknown. This study aimed to identify ageing-related biomarkers in OA patients with MetS. OA and MetS datasets and ageing-related genes (ARGs) were retrieved from public databases. The limma package was used to identify differentially expressed genes (DEGs), and weighted gene coexpression network analysis (WGCNA) screened gene modules, and machine learning algorithms, such as random forest (RF), support vector machine (SVM), generalised linear model (GLM), and extreme gradient boosting (XGB), were employed. The nomogram and receiver operating characteristic (ROC) curve assess the diagnostic value, and CIBERSORT analysed immune cell infiltration. We identified 20 intersecting genes among DEGs of OA, key module genes of MetS, and ARGs. By comparing the accuracy of the four machine learning models for disease prediction, the SVM model, which includes CEBPB, PTEN, ARPC1B, PIK3R1, and CDC42, was selected. These hub ARGs not only demonstrated strong diagnostic values based on nomogram data but also exhibited a significant correlation with immune cell infiltration. Building on these findings, we have identified five hub ARGs that are associated with immune cell infiltration and have constructed a nomogram aimed at early diagnosing OA patients with MetS.
1 Dec 2025
Read appraisal →