Spatially resolved contrastive analysis of perturbation-driven microenvironmental heterogeneity across tissues and conditions

Abstract

Understanding how tissues remodel in response to perturbations requires computational tools that can untangle condition-specific changes from the conserved tissue architecture. We present Haruka, a spatially aware contrastive learning framework that identifies salient (condition-specific) and background (shared) spatial domains across tissue slices and experimental conditions. Haruka integrates contrastive variational inference with an auxiliary microenvironment reconstruction task, enabling the model to learn spatial-context-informed embeddings that capture both perturbation effects and local neighborhood context. Benchmarking on simulated and real datasets demonstrates superior detection of spatially heterogeneous perturbation responses compared to existing methods. Applied to diverse spatial omics platforms, Haruka distinguished immunotherapy responders in melanoma, traced fibrosis progression in human lung tissue, and mapped treatment-resistant microenvironments in KRASG12D-mutated lung cancer. Extended to spatially resolved multiomics data co-profiling transcriptomics and chromatin accessibility from an LPC-induced neuroinflammation mouse model, using a modality-parallel design that independently models each omics layer before integration, Haruka resolved condition-specific regulatory programs encoded across chromatin and transcriptome layers, linking spatial domain identity to transcription factor network architecture and human psychiatric disease genetics. Thus, Haruka provides a generalizable framework for spatial contrastive analysis, enabling systematic dissection of tissue organization, cellular plasticity, and microenvironmental remodeling across disease contexts and therapeutic perturbations.