Advances in single cell biology using ultra-high throughput data

Jihwan Park

Gwangju Institute of Science and Technology

The recent advances in ultra-high-throughput single-cell sequencing technologies have fueled an exponential growth of single-cell sequencing data available in the public, which currently harbors hundreds of millions of cells in total. Using the public data, we performed pan-cancer analysis on 226 samples from 164 donors across 10 types of solid cancers to profile tumor microenvironment (TME) at single-cell resolution. The activation trajectory of fibroblasts into various major types of cancer-associated fibroblasts was divided into three distinct differentiated states, and was capable of sculpting TME, which was associated with prognosis of immunotherapy. However, currently available bioinformatic tools for single cell analysis, including SCANPY, Seurat, and Monocle, are largely inappropriate for the integrative analysis of large single cell data containing millions of cells due to the extensive memory footprint required by such analysis. Here, we developed SC-Elephant, an extremely memory-efficient single cell analysis platform that enables routine analysis of millions of cells using a typical desktop or billions of cells in a server.