Identify the origin and end cells and infer the trajectory of cellular fate automatically

Abstract

Cellular state identification and trajectory inference enable reconstructions of cell fate dynamics from single-cell RNA sequencing. However, the identification of cell fate trajectories requires a large number of computational resources or known biological process, and lack a method to alleviate both of these deficiencies at the same time. Here, we present scLTNN, a method that automatically infers origin and end cell state from scRNA-seq data and calculates the developmental trajectory and differentiation direction of cells with only a few computational resources and time consummation. We apply scLTNN to disentangling subpopulation kinetics in CD8+ T cell, pancreatic endocrinogenesis, and the development of zebrafish embryos. scLTNN displays a strong trajectory inference ability cross-species. scLTNN features a modular design that can be flexibly extended to any scRNA-seq analysis task. The complete package is available online at https://github.com/Starlitnightly/scltnn.