The advent and rapid adoption of single cell RNA sequencing technology have ushered in an era of biological breakthroughs at the cellular level. Despite the biological and technical variation in addition to computational challenges, the ability to isolate individual cells and generate their sequencing libraries have enabled researchers to investigate topics such as the discovery of cell sub-populations and ability to infer transcriptional dynamics. On this note, we will explore the latent representations of the single cell RNA sequencing data expression matrix to investigate two complex processes: clonal hematopoiesis and circadian rhythms.

For clonal hematopoiesis, we examine the inflammatory response of a patient with myeloproliferative neoplasms (MPN). Here, we present a computational approach that first identifies ranked groups of differentially expressed genes and then uses those groups to cluster the hematopoietic stem and progenitor cells (HSPCs). We confirm that the MPN patient shows more response to inflammation than the unaffected patient. For circadian rhythms, we present a new framework that accurately predicts the circadian time of transcriptomics samples. We show that this framework can aid in the development of circadian precision medicine disease management plans and that it can also provide evidence of circadian heterogeneity in different cell types.