There has been enormous progress in single-omics spatial technologies, which has revolutionized our understanding of the tumor microenvironment (TME). While the data generated from these methods have helped to unravel cellular intricacies within spatial context at the genomic, transcriptomic, metabolomic and protein levels, users are increasingly combining different omics readouts to obtain a holistic view of TME heterogeneity and complexity. However, spatial multi-omics data analysis presents specific bioinformatics challenges, as the data is typically acquired at different spatial resolutions, using a variety of platforms, and generates large data volumes. We present Weave, a cloud-based software for spatial omics bioinformatics, enabling efficient integration and joint visualization of different spatial-omics assays.

Human lung cancer sections were sequentially analyzed with spatial transcriptomics using a cancer panel targeting 289 genes (Xenium, 10X Genomics), followed by multiplexed immunofluorescence using a 40-antibody panel (COMET, Lunaphore), and then H&E staining. The H&E images were digitized (Axioscan 7, Zeiss), and pathology annotation performed in QuPath. Cell segmentation was performed on the Xenium dataset using DAPI-based nuclear expansion (10X Genomics), and on the COMET dataset using CellSAM. All data were co-registered at full resolution using a non-rigid spline-based algorithm, then visualized in a web-based viewer (Weave, Aspect Analytics).

Weave software was developed to address several spatial omics bioinformatics challenges. This software fully supports joint visualization of different spatial omics assays, and multiple common downstream multimodal analysis pipelines. The cloud-based software allows interactive browsing of datasets at full resolution via web-browsers, enabling communication of results between collaborators, and removes limitations of location or operating system. For the use case presented here, we utilized an advanced integration pipeline to match readouts across the COMET, Xenium and H&E data, accounting for different sized measurement regions and spatial resolutions. Pathology annotations and cell segmentation results were integrated and overlaid as additional visualization layers. We conducted correlation analysis to identify which transcript-protein pairs had similar spatial expression and if this correlation was affected by cell segmentation approach. Some pairs had high correlation regardless of cell segmentation, while proteins from complexes derived from multiple genes (e.g. CD3) yielded variable correlations.

As spatial multi-omics is increasingly used to investigate TME biology, we present software addressing the need for appropriate bioinformatics solutions.