Uncovering disease-related multicellular pathway modules on large-scale single-cell transcriptomes with scPAFA

Pathway analysis is a crucial analytical phase in disease research on single-cell RNA sequencing (scRNA-seq) data, offering biological interpretations based on prior knowledge. However, currently available tools for generating cell-level pathway activity scores (PAS) exhibit computational inefficacy in large-scale scRNA-seq datasets. Besides, disease-related pathways are commonly identified by cross-condition comparisons in each cell type, neglecting the potential multicellular patterns. Here, we present single-cell pathway activity factor analysis (scPAFA), a Python library designed for large-scale single-cell dataset allowing rapid PAS computation and uncovering biologically interpretable disease-related multicellular pathway modules, which are low-dimensional representations of disease-related PAS variance in multiple cell types. Application on colorectal cancer (CRC) dataset with 371,223 cells and large-scale lupus atlas over 1.2 million cells demonstrated that scPAFA can achieve > 33-fold decreases in runtime of PAS computation and further identified reliable and interpretable multicellular pathway modules that capture the transcriptomic features of CRC tumor status and transcriptional abnormalities in lupus patients, respectively.