Yang, Chen

Abstract Coarsening Visium HD resolution from 8 to 64 µ m can flip cell-type co-localization from negative to positive ( r = −0.12 → +0.80 ), yet many widely used compositional deconvolution workflows require coarsening or subsampling at million-bin scale. Here we introduce FlashDeconv, which combines leverage-score importance sampling with sparse spatial regularization to achieve competitive benchmark accuracy while processing 1.6 million bins in 153 seconds on commodity hardware. Systematic multi-resolution analysis of Visium HD mouse intestine reveals a tissue-specific resolution horizon (8–16 µ m)—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 largely missed by discrete-label summaries, with RCTD doublet mode labeling only 2.3% of hotspot bins as neutrophil singlets.

Abstract Different large language models (LLMs) have the potential to complement one another. We introduce an iterative multi-LLM consensus framework for annotating single-cell RNA sequencing data. This framework outperforms the best state-of-the-art method by nearly 15% in mean accuracy (77.3% vs 61.3%) across 50 diverse datasets from 26 tissues, encompassing over 8 million cells. By leveraging cross-model deliberation, our framework quantifies uncertainty, identifies ambiguous clusters for expert review, provides transparent reasoning chains, and minimizes the effort and expertise needed for cell type annotation in large-scale studies. Additionally, our framework enables users to seamlessly integrate new LLMs.