Identification of critical cell-types using genetic modules: A case study of neurodevelopmental disorders

Abstract

Identifying the critical cell-types impacted by various diseases is crucial for understanding disease mechanisms and developing targeted therapeutics. Recent advances in disease genetic module discovery and single-cell technologies provide a unique opportunity to study critical cell-types based on functional pathways and modules. Disease genetic modules are defined as sets of genes with correlated expression that are part of the same biological pathways and are disrupted in the disease. Critical cell-types for a biological function are defined as clusters of similar cells most "active" or "involved" in that biological function. In this paper, we provide a formal problem definition for the critical cell discovery problem using the recently introduced local correlation concept, and show that the proposed problem is intractable in theory. We propose a novel method, MoToCC (Module To Critical Cell-types), to find sets of similar cells with local correlated gene expression activity for input modules. We evaluated MoToCC on four neurodevelopmental disorder modules using single-cell expression data from the developing human cortex. Finally, we demonstrate that the objective value returned by MoToCC for the tested modules is an acceptable approximation to the optimal solution. Overall, our work provides a valuable tool for studying critical cell-types and their role in disease mechanisms, which could lead to the development of more effective targeted therapeutics. The MoToCC package is available at https://github.com/jchow32/MoToCC