A strong week for population-scale single-cell genetics, plus new reference-mapping and spatial tools. Scope read as single-cell and spatial omics methods and atlases; for several items only abstracts or excerpts were accessible, noted below. No funding calls were verified on official pages this week.
01Nature · Sep 30Population-scale immune multiome atlas connects disease variants to mechanism
The paper reports simultaneous single-cell chromatin accessibility and gene expression profiling across immune cells to map molecular QTLs and connect variants to disease. According to the Broad-led press release, the work covers millions of immune cells from more than 1,100 blood donors in FinnGen, with collaborators at BioBank Japan; the abstract describes validating 10,428 fine-mapped molecular QTLs.
The authors report that at evolutionarily constrained genes, accessibility changes have normal effect sizes but are attenuated on the way to expression via weaker, more numerous enhancer–gene links. I read the abstract excerpt and a press release (not the full text), so treat the effect claims as the authors' own.
For researchers: if you fine-map immune GWAS loci, check whether the atlas's joint accessibility-expression QTLs resolve your variants better than eQTL-only references.
02bioRxiv · Sep 30scTAPAS reconstructs donor genotypes from scRNA-seq reads for eQTL mapping
Preprint (not peer-reviewed). The authors present scTAPAS, a framework that reconstructs genotypes directly from 5' scRNA-seq reads using reference-based imputation, tested on the COVID-19 Multi-omic Blood Atlas (COMBAT).
They report that, testing only 18.0% of the variants available from array genotyping plus imputation, it recovers 68.6% of cell type–eGene pairs, with strong effect-size concordance. They also report CD4+ T cell-specific associations between HLA class II alleles and TCR Vα usage. Only the abstract was read.
For researchers: this could make single-cell eQTL analysis feasible on older cohorts lacking genotypes; validate on your data, since results come from one cohort.
03Nature Genetics · Sep 30ArchMap offers web-based reference mapping of query datasets to atlases
This open-access Brief Communication describes ArchMap, a web platform for reference-based analysis of single-cell datasets. As a demonstration, the authors mapped an idiopathic pulmonary fibrosis dataset to the Human Lung Cell Atlas.
They report 96.85% gene overlap, full Leiden cluster preservation (5 of 5), 53.96% of query cells with a reference anchor in a similar state, and 8.99% flagged 'unknown' (uncertainty >0.5). Disease tissue gave fewer anchors (54%) than healthy tissue (63%), as the authors note.
For researchers: try ArchMap for quick atlas annotation, but expect lower anchor rates on diseased samples and review the 'unknown' fraction.
04Nature Methods · Sep 30CytoVI brings probabilistic modelling to cytometry and CITE-seq-style protein data
CytoVI is a single probabilistic model that generates cell embeddings, imputes missing measurements, tests differential protein expression and automates cell annotation for antibody-based single-cell data.
The authors applied it to build an integrated B cell maturation atlas spanning 350 proteins and to find T cell states associated with nodal B cell lymphomas. Abstract and main-text excerpts were read; the CITE-seq confirmation was noted by the authors as underpowered.
For researchers: if you integrate cytometry panels across cohorts or technologies, CytoVI is a candidate to benchmark against your current batch-correction workflow.
05Nature Comput. Sci. · Oct 5CellART unifies segmentation and cell typing for subcellular-resolution spatial data
This Brief Communication introduces CellART, which combines staining images, spatial transcriptomics data and scRNA-seq references to perform cell segmentation and cell-type annotation together, using deep learning and probabilistic modelling.
The authors describe it as efficient, generalizable and robust across high-resolution platforms, with outputs compatible with community tools. The article is paywalled; only the abstract was read, and no benchmark numbers were verified.
For researchers: worth testing on Visium HD or imaging-based data, but wait for the full text's benchmarks before replacing your segmentation pipeline.
06Keystone · Oct 21Keystone single-cell meeting: short-talk abstracts and scholarships due Oct 21
Keystone Symposia's 'Single Cell Biology: Expanding Potential Across Diseases' runs Feb 15–18, 2027 in Banff, Canada, joint with Tissue and Spatial Immunology.
The official page lists Oct 21, 2026 for both the short talk abstract and scholarship deadlines, Dec 16, 2026 for early registration and Jan 25, 2027 for poster abstracts. No time zone is stated on the page.
For researchers: diary Oct 21 if you want a short talk or scholarship; confirm the cutoff time on the Keystone site.
Enable JavaScript to read this edition.