Quantitative Learning of Cellular Features From Single-cell Transcriptomics Data Facilitates Effective Drug Repurposing

Abstract

In this study, we have devised a computational framework SuperFeat that allows for the training of a machine learning model and evaluate the canonical cellular states/features in pathological tissues that underlie the progression of disease. This framework also enables the identification of potential drugs that target the presumed detrimental cellular features. This framework was constructed on the basis of an artificial neural network with the gene expression profiles serving as input nodes. The training data comprised single-cell RNA-seq datasets that encompassed the specific cell lineage during the developmental progression of cell features. A few models of the canonical cancer-involved cellular states/features were tested by such framework. Finally, we have illustrated the drug repurposing pipeline, utilizing the training parameters derived from the adverse cellular states/features, which has yielded successful validation results both in vitro and in vivo. SuperFeat is accessible at https://github.com/weilin-genomics/rSuperFeat.