SpaceNet: Pathology‐guided analysis of lesion‐associated spatial molecular programs

Disease-associated molecular alterations are often confined to localized pathological regions, yet current spatial transcriptomic analyses do not always use pathology or imaging information as the primary reference for defining disease-relevant spatial contexts. We present SpaceNet, a pathology-guided analytical framework integrating pathology or imaging data with spatial transcriptomics and single-cell/single-nucleus RNA sequencing to characterize localized cellular and molecular organization. SpaceNet defines regions of interest (ROIs) from pathology- or imaging-derived features and integrates cell-type composition, gene expression, interaction information, and ROI-restricted spatial co-expression to infer pathology-associated molecular networks. Benchmark analyses showed that pathology-guided ROI definition improved localization of lesion-associated regions and the recovery of disease-related cellular and molecular features. Applications to multimodal Alzheimer's disease datasets further demonstrated SpaceNet's ability to characterize complementary pathology-associated spatial microenvironments across imaging modalities. SpaceNet provides a flexible and biologically interpretable framework for analyzing disease-associated spatial organization.