Consensus tissue domain detection in spatial multi-omics data using MILWRM

Abstract

Spatially resolved molecular assays provide high dimensional genetic, transcriptomic, proteomic, and epigenetic information in situ and at various resolutions. Pairing these data across modalities with histological features enables powerful studies of tissue pathology in the context of an intact microenvironment and tissue structure. Increasing dimensions across molecular analytes and samples require new data science approaches to functionally annotate spatially resolved molecular data. A specific challenge is data-driven cross-sample domain detection that allows for analysis within and between consensus tissue compartments across high volumes of multiplex datasets stemming from tissue atlasing efforts. Here, we present MILWRM (multiplex image labeling with regional morphology) a Python package for rapid, multi-scale tissue domain detection and annotation. We demonstrate MILWRM's utility in identifying histologically distinct compartments in human colonic polyps and mouse brain slices through spatially- informed clustering in two different spatial data modalities. Additionally, we used tissue domains detected in human colonic polyps to elucidate molecular distinction between polyp subtypes. We also explored the ability of MILWRM to identify anatomical regions of mouse brain and their respective distinct molecular profiles.

Molecular cartography uncovers evolutionary and microenvironmental dynamics in sporadic colorectal tumors

“…Application of cell-cell interaction community reconstruction algorithms to ST data provided inaccurate results due to low spatial resolution and large distances between microwells. Focusing on pixel-based community detection, we employed refNMF usages as predictors for a MILWRM model to divide tissue into consensus domains based on cell-state makeup 63 (Figure 4D). The MILWRM model yielded eight domains (D0-D8) that correspond to CIN+ epithelium (D4 - high in CRC2 and STM), normal mucosa (D5 - enriched in ABS and CT), and sessile-serrated epithelium (D0 - high in SSC and GOB), amongst others, all of which spatially align with regional histology (Figure 4A-E; Figure S4H).…”

“…Tissue domain detection with MILWRM. We employed refNMF cell states as predictors for a MILWRM model of macro-level consensus tissue domains across all ST slides 63 . We only included states from epithelial and stromal compartments (Figure 4G), excluding immune states to limit predictors to markers of tissue architecture.…”