Regulatory Driven Clustering of Single-Cell Data; Clustering of single-cell RNA sequencing from glioblastoma with a novel mathematical method

Abstract

Cancer is a leading cause of death worldwide. Single-cell RNA sequencing has arisen as an important method to explore the gene expression of biological cells, including cancer cells. In this study, we deployed a computational algorithm known as ScRegClust to dissect single-cell RNA-sequencing (scRNA-seq) data from brain tumors. This method uncovers modules of co-expressed genes, and identifies corresponding regulators, such as transcription factors and kinases. We sought to discern whether distinct scRNA-seq datasets could mutually inform each other by examining the patterns of gene clustering and regulatory mechanisms. The goal was to leverage this knowledge to guide the algorithm in a subsequent run, thereby enhancing its performance. Although the preliminary findings from simulated data offered promising prospects, transitioning to real-world data consisting of glioblastomas presented considerable hurdles. While our results shed light on the intricacies of reconstructing regulatory programs, the overall performance did not meet our initial projections. These findings underscore the complexity of and challenges associated with scRNA-seq analysis, underscoring the necessity for further exploration and refinement of current methodologies. This research enriches the field of data integration in cancer genomics and lays a foundation for future efforts aimed at refining regulatory-driven clustering of single-cell data.