Abstract
Twenty or more spatial-transcriptomics deconvolution algorithms have been published with inconsistent benchmark results. None has been derived from formal signal-separation theory. This Opinion frames deconvolution as a blind source separation problem, derives identifiability conditions, maps existing algorithms to their BSS equivalents, and proposes principled algorithm selection criteria based on formal signal-model assumptions.
Venues
Preprint: bioRxiv · Bioinformatics
Journal target: Briefings in Bioinformatics (Oxford University Press)
Citation
Walker A. (2026). Spatial Transcriptomics Cell Type Deconvolution as a Blind Source Separation Problem: A Formal Signal Processing Framework. Vitacore Research Collaborative. bioRxiv · Bioinformatics.