Team builds a single-nucleus multi-omic atlas across 21 adult human tissues, profiling 459,856 transcriptomic and chromatin accessibility profiles and identifying 161,270 novel regulatory elements

The study builds a single-nucleus multi-omic atlas covering 21 adult tissues and four donors, comprising 459,856 transcriptomic and chromatin accessibility profiles, with paired measurements from 160,688 nuclei, resolving nine cell lineages, 61 broad cell types and 313 subclusters. Compared with previous multi-omic reference maps, this work measures transcriptome and chromatin accessibility within the same nucleus and extends coverage to 21 adult tissues, providing more systematic cross-tissue cell-type resolution. Based on 459,856 profiles and 160,688 paired nuclei across 21 tissues and four donors, the scale and paired multi-omic design provide direct evidence for cell-type resolution.

The study identifies 1,085,062 candidate cis-regulatory elements, including 161,270 novel elements absent from ENCODE, and through joint profiling obtains 871,177 cCRE-gene associations, revealing lineage-specific regulatory architectures. The 161,270 novel elements expand the known regulatory element catalog, and the 871,177 cCRE-gene associations directly connect regulatory DNA to cellular expression, providing a cross-tissue reference for interpreting non-coding genetic risk. Joint analysis of paired multi-omic data yields cCRE-gene associations, and novel elements are determined by comparison with the ENCODE catalog, with evidence from large-scale cross-tissue data.

Cross-tissue accessibility analysis identifies lineage-restricted and constitutively inaccessible chromatin domains, the latter showing preferential hypomethylation across human cancers. The work links chromatin accessibility states to cancer methylation patterns, suggesting that constitutively inaccessible regions may carry epigenetic features across cancer types. Observational association based on cross-tissue accessibility comparisons and cancer methylation data, with evidence from cross-tissue analysis of the atlas.

The study uses the dataset to train sequence-to-function models that predict chromatin-accessibility effects for 548,656 fine-mapped variants, identifying 18,133 high-effect variants including 1,120 broadly active variants, and trains models for eight endothelial subtypes to resolve variant effects across vascular beds. The work converts atlas data into computational models that predict variant regulatory effects and refines predictions for endothelial subtypes, providing a computational framework for complex trait genetics interpretation. Based on predictions for 548,656 fine-mapped variants and identification of 18,133 high-effect variants, with model training and prediction grounded in the atlas data and endothelial subtype models covering eight subtypes.

The atlas is intended for regulatory sequence interpretation and variant effect prediction within the scope of 21 adult tissues and four donors, primarily for researchers in functional genomics, genetics and computational biology. The sequence-to-function models can be used to predict chromatin-accessibility effects for fine-mapped variants, and the endothelial subtype models apply to resolving variant effects across vascular beds. This resource provides a reference framework for subsequent cross-tissue regulatory studies and complex trait genetics analyses.

The current text is an abstract and competing interest statement, lacking figures, supplementary materials and full methodological details, so data quality control, model validation strategies and independent validation results cannot be assessed. Readers may watch for: whether paired measurements are evenly distributed across the 21 tissues, functional validation of novel regulatory elements, generalization of sequence-to-function models across populations and cell types, and mechanistic explanations for the association between constitutively inaccessible regions and cancer hypomethylation.