Systematic evaluation of single-cell multimodal data integration for comprehensive human reference atlas

We systematically evaluated experimental and computational strategies for integrating multimodal single-cell and single-nucleus genomics data in complex human tissues. Using human renal cortex as the primary tissue, we generated a benchmarking dataset comprising 3' and 5' single-cell RNA sequencing (scRNA-seq) as well as paired single-nucleus RNA sequencing and chromatin accessibility profiling (snRNA-seq and snATAC-seq). Samples were obtained from healthy adult donors undergoing nephrectomy, resulting in 33 kidney cortex samples from 19 donors. In total, over 119,000 high-quality cells and nuclei were profiled. The study evaluates horizontal, vertical, and global multimodal integration strategies and introduces an interpretable machine-learning framework (scOMM) to harmonize... (for more see dbGaP study page.)