VIMA discovers disease-associated tissue structures without cell-type annotation

Summary

Nature Methods, Published online: 06 October 2026; doi:10.1038/s41592-026-03242-3 Identification of disease-associated patterns in spatial molecular data is challenging. We introduce variational inference-based microniche analysis (VIMA), a deep learning-based statistical method that can identify such patterns without requiring annotation of the data into cell types or niches. VIMA has high power and fidelity across a range of spatial molecular technologies and diseases.