Scalable, Generalizable, and Uncertainty-Aware Integration of Spatial Multi-Omics Across Diverse Modalities and Platforms with SCIGMA

Recent advances in spatial omics technologies have enabled simultaneous profiling of transcriptomic, proteomic, epigenomic, metabolomic, and imaging data at high spatial resolution, offering unprecedented opportunities to dissect tissue complexity. However, integrating these diverse and large-scale spatial multi-modal datasets remains a major computational challenge. We present SCIGMA, a scalable and generalizable deep learning framework for spatial multi- omics integration. SCIGMA introduces a novel uncertainty-aware contrastive learning objective and multi-view graph neural networks to preserve modality-specific signals while learning biologically meaningful joint representations. Unlike existing methods, SCIGMA provides spatially resolved uncertainty estimates, improving interpretability and identifying regions of biological or technical heterogeneity. SCIGMA is the first spatial multi-omics method to support integration of up to five modalities - as demonstrated on Spatial-Mux-Seq data - and its modular framework is extensible to future technologies with even more modalities. It also scales to over one million spatial locations, enabling analysis of ultra-high-resolution datasets such as VisiumHD and Xenium Prime. We evaluated SCIGMA across 19 datasets spanning 8 modalities, 10 tissues, and 9 platforms. On benchmarkable datasets, SCIGMA outperformed existing methods in spatial domain detection, modality preservation, feature reconstruction, and reproducibility. Across applications, it uncovered biologically meaningful structures, refined spatial domains, and modality-specific regulatory programs, while its uncertainty estimates revealed tissue regions with potential biological or technical variation. Together, SCIGMA provides a robust, flexible, and future-ready solution for scalable spatial multi-modal integration.