SciGeneX: Enhancing transcriptional analysis through gene module detection in single-cell and spatial transcriptomics data.

Abstract The standard pipeline to analyze scRNA-seq or spatial transcriptomics data focuses on a gene-centric approach, which overlooks the collective behavior of genes. However, cell populations should be viewed as intricate combinations of activated and repressed pathways. Thus, a broader view of gene behavior would provide more accurate information on cellular heterogeneity in single-cell or spatial transcriptomics data. Here, we described SciGeneX, a R package implementing a neighborhood analysis and a graph partitioning method to generate co-expression gene modules. These gene modules, which can be shared or restricted between cell populations, collectively reflect cell populations, and their combinations are able to highlight specific cell populations, even rare ones. SciGeneX was also able to uncover rare and novel cell populations which were not observed before in spatial transcriptomics data of human thymus. We show that SciGeneX outperforms existing methods on both artificial and experimental datasets. Overall, SciGeneX will aid in unraveling cellular and molecular diversity in single-cell and spatial transcriptomics studies. The R package is available at https://github.com/dputhier/scigenex .