Jimba, T. | bioRxiv Science

bioRxiv2026

Multimodal Hypersampling of the adult human heart resolves tissue-specific cell states and disease-associated variants

Cardiovascular diseases often arise in anatomically specialised regions of the heart that remain poorly represented in existing single-cell and spatial atlas references. Here we profiled 28 regions of the adult human heart from 36 donors, combining paired single-nucleus RNA and chromatin accessibility assays with spatial transcriptomics. The resulting atlas integrates 1,059,175 expression profiles and 452,614 chromatin accessibility profiles, and defines spatial niches across the heart. In cardiac valves, we identify a valve fibroblast-macrophage niche that is polarised to the non-fibrosa surface and expands with age. Aortic stenosis genetic risk was enriched in inflammatory regulatory programmes of the Valve Fibroblast Immune cell state. In pulmonary veins, we define PITX2+ Myocardial Sleeve Cells, a region-restricted cardiomyocyte population in which a fine-mapped 4q25 atrial fibrillation variant intersects a cell type-specific open chromatin region. In coronary arteries, we resolve a RUNX1-marked synthetic smooth muscle layer in the subintima that provides a molecular identity for diffuse intimal thickening, a constitutive feature of the healthy vessel wall and a substrate for atherosclerosis, and show by perturbation in human smooth muscle cells that RUNX1 promotes cell-cycle entry while restraining inflammatory cytokine gene expression. Finally, allele-specific accessibility together with deep learning models fine-tuned to predict regulatory activity from DNA sequence identify cell type-specific regulatory effects at GWAS loci for atrial fibrillation, coronary artery disease and calcific aortic valve stenosis, including a variant whose Valve Fibroblast-specific effect on the polyamine transporter ATP13A3 is opposite in direction to the effect seen in bulk-tissue eQTL references that lack valve cells. Together with these cardiac sequence models and CardioSleuth, an AI-powered exploration engine for cardiac regulatory effects, this atlas links cardiac anatomy, gene regulation and disease genetics at single-cell resolution. By placing genetic risk in its precise cellular and anatomical context, it provides a foundation for mechanistic disease understanding, cell-state-specific therapeutic target nomination, and improved interpretation of inherited cardiovascular risk to inform future therapy and management.

Cell Biology↗

bioRxiv2024

inVAE: Conditionally invariant representation learning for generating multivariate single-cell reference maps

Single-cell data is driving new insights into the spatiotemporal dynamics of cells and individual disease susceptibility. However, accurately identifying cell states across diverse cohorts remains challenging, as both biological variation and technical biases cause distributional shifts in the data. Separating these effects is crucial for capturing cellular heterogeneity and ensuring interpretability. To address this, we developed inVAE, a conditionally invariant deep generative model based on variational autoencoders. inVAE models the latent space as a combination of invariant variables, encoding true biological signals, and spurious variables, capturing technical biases. By conditioning the prior distribution of cells on biological covariates, such as disease variants, inVAE identifies high-resolution cell states in the invariant representation. Enforcing independence between the two representations disentangles biological signals from noise, enabling a more interpretable and generalizable model with a causal semantic. inVAE outperformed existing methods across four human cellular atlases of the human heart and lung, while uncovering novel cell states. It precisely stratified cell atlas donors based on the genetic impact of pathogenic variants, and excelled in predicting cell types and disease in unseen data, proving its generalizability as a reference model for label transfer. Furthermore, inVAE accurately identified temporal cell states and trajectories from developmental datasets, and captured spatial cell states in a spatially-resolved atlas. In summary, inVAE provides a powerful method for integrating multivariate single-cell transcriptomics data. By leveraging prior knowledge such as metadata, it effectively accounts for biological variation and improves latent space interpretability by disentangling biological and technical sources of variation. These capabilities enable deeper insights into cellular heterogeneity and its role in disease progression.

bioinformatics↗