Hastings: An R pipeline for large-scale RNA-Seq data analysis

Abstract

Motivation: With the advance of high-throughput sequencing technologies, large-scale datasets becomes increasingly common, from thousands of patients involved in a cohort study to thousands of single cells from the same tissue. Despite major advances in understanding the molecular mechanisms governing biological processes from development to diseases, heterogeneity from between two individual persons to two cells from the same tissue is yet to be scrutinized. Several studies have revealed new insights from such large-scale datasets [1, 2], yet their analysis protocols are not always straightforward to be adopted or extend. In particular, it is often more of an art to select various parameters along the whole analysis procedures, and visualization of high dimensional data could be very challenging. Results: Here, we present Hastings to face the demand of large-scale data analysis and visualization for RNA-Seq gene expression data. As demonstrated in the three examples, Hastings can eficiently identify sub-groups in an unsupervised manner, identify potential marker genes and generate clear 2D visualizations. Hastings could be widely applied from bench to clinics.