FlashDeconv enables atlas-scale, multi-resolution spatial deconvolution via structure-preserving sketching

Coarsening Visium HD resolution from 8 to 64 µm can flip cell-type colocalization from negative to positive (r = −0.12 → +0.80), yet investigators are routinely forced to coarsen because current deconvolution methods cannot scale to million-bin datasets. Here we introduce FlashDeconv, which combines leverage-score importance sampling with sparse spatial regularization to match top-tier Bayesian accuracy while processing 1.6 million bins in 153 seconds on a standard laptop. Systematic multi-resolution analysis of Visium HD mouse intestine reveals a tissue-specific resolution horizon (8–16 µm)—the scale at which this sign inversion occurs—validated by Xenium ground truth. Below this horizon, FlashDeconv provides, to our knowledge, the first sequencing-based quantification of Tuft cell chemosensory niches (15.3-fold stem cell enrichment). In a 1.6-million-bin human colorectal cancer cohort, FlashDeconv uncovers neutrophil inflammatory microdomains co-localized with immunoregulatory dendritic cells (mRegDC) at the tumor–stroma interface—spatial niches invisible to classification-based methods, which discard 97.7% of the relevant bins. Competing Interest Statement The authors have declared no competing interest. Footnotes Contributing authors: cafferychen777{at}tamu.edu;

rewritten to emphasize quantitative results (resolution-dependent sign inversion, processing speed, Tuft cell niche quantification, and neutrophil microdomain discovery). Introduction streamlined: condensed the leverage-vs-variance exposition and moved detailed derivations to Methods/Supplementary. New case study: replaced the HGSOC treatment-response application with an atlas-scale human colorectal cancer (CRC) Visium HD analysis (1.6 million bins at 8-micron resolution), including tumor-stroma boundary immune gradients and neutrophil inflammatory microdomains co-localized with immunoregulatory dendritic cells (mRegDC). Added cross-platform validation using Xenium in situ sequencing from an adjacent serial section, demonstrating Pearson r > 0.88 across resolutions from 8 to 128 microns. Expanded Discussion with three failure-mode analysis (feature-space collinearity in melanoma, signal-space rare markers in intestine, and measurement-space continuous mixing in CRC) and biological interpretation of neutrophil-mRegDC co-localization in the context of tumor immunity. Revised Tuft-Stem niche section: added raw gene expression validation (Pou2f3-Lgr5 neighbor enrichment), cross-method comparison with RCTD, and clarified that FlashDeconv quantifies rather than discovers this spatial architecture. Updated benchmark presentation: combined accuracy and scalability into a single figure; added RCTD comparison throughout. Figure panel labels standardized to lowercase. Supplementary materials updated with new CRC analyses, Xenium validation, and expanded methodological notes. https://github.com/cafferychen777/flashdeconv-reproducibility