Spatial multi-omics imputation and embedding with SpaMIE

Clinical Snapshot

Key Findings

• P Value: Not reported

• Effect Size: Not reported in abstract — quantitative performance metrics (e.g., ARI, RMSE, correlation coefficients) are not provided

• Primary Outcome: SpaMIE achieves accurate cross-modal imputation of missing spatial omics modalities, robust multi-section integration, and improved spatial domain identification compared to existing methods, as demonstrated on simulated and experimental SMO datasets

• Nnt Or Sensitivity: Not applicable to this computational methods study; analogous metrics (e.g., imputation correlation, clustering ARI improvement) are not quantified in the abstract

• Confidence Interval: Not reported

Clinical Application

Feasibility depends on availability of computational infrastructure (GPU resources for deep learning), bioinformatics expertise to implement and tune the GNN framework, and access to at least some fully profiled multi-omics sections to train the imputation model. The tool is described as scalable, but runtime and resource requirements are unquantified. Open-source availability is implied but not confirmed in the abstract. Australian research institutions with spatial multi-omics capabilities — including those within the EMBL Australia network, Garvan Institute, Walter and Eliza Hall Institute, and university-based genomics centres — stand to benefit from SpaMIE if it enables cost-effective atlas construction. The tool is not subject to TGA regulation (it is a research software tool, not a medical device or diagnostic). PBS and RACGP guidelines are not directly relevant. However, Australian researchers contributing to international spatial atlas consortia (e.g., Human Cell Atlas) may find SpaMIE valuable for integrating heterogeneous datasets. Adoption would require local bioinformatics capacity and access to HPC or cloud computing resources, which vary across institutions. Research laboratories and bioinformatics teams constructing large-scale spatial multi-omics atlases from tissue sections where full multi-omics profiling of every section is cost-prohibitive. Relevant to translational research in oncology, neuroscience, developmental biology, and any field utilising spatial transcriptomics or spatial proteomics platforms.

Abstract

SpaMIE is a deep graph neural network framework designed to tackle the challenge of multi-section integration in spatial multi-omics (SMO) datasets with systematic missing modalities. Current SMO platforms face limitations such as high cost and limited throughput, leading to many large-scale spatial atlases relying on cost-effective mono-omics measurements while only a few sections are profiled with full multi-omics technologies. This results in heterogeneous modality coverage across tissue sections. SpaMIE offers a two-stage solution. In the first stage, it performs spatially informed cross-modal imputation, enabling accurate inference of missing modalities from mono-omics data. In the second stage, it integrates measured and imputed spatial multi-omics profiles across multiple tissue sections to learn a unified embedding. Benchmarking on simulated and experimental datasets shows that SpaMIE achieves accurate cross-modal imputation, robust multi-section integration, and improved spatial domain identification, providing a flexible and scalable solution for constructing and analyzing SMO atlases.

References

1. 1.Liu, W., Xiang, D., Jiang, X., Bai, Y., Li, S., Li, W., Lv, S., Li, Q., Jiang, J., & Liu, J. (2026). Spatial multi-omics imputation and embedding with SpaMIE. Cell Reports Methods. https://doi.org/10.1016/j.crmeth.2026.101456