Approaches to Marker Gene Identification from Single-Cell RNA-Sequencing Data

Ronnie Y. Li, Wenjing Ma, Zhaohui S. Qin

Abstract: The advent of next-generation sequencing techniques, such as single-cell RNA-sequencing (scRNA-seq), has drastically increased the amount of data generated and available for analysis. One significant problem for scRNA-seq is identifying and characterizing highly heterogeneous cell types. Many approaches have been devised to tackle this challenge, mostly utilizing the genetic expression patterns of cells to infer cell type. Research hopes to identify “marker genes” that exhibit cell-type-specific expression levels, thus providing a way to distinguish one group of cells from the others. There are several marker gene identification and selection tools available; however, each method has strengths and limitations, making it difficult to choose the optimal marker gene identification method for a given dataset. In addition, recent computational methods can suffer from long training times, and they often rely on laborious manual annotations. Using recent literature, we assessed several well-known marker gene selection methods, including COMET, RankCorr, scGeneFit, Seurat, SC3, SCMarker, and scTIM. The abundance of these tools illustrates the complexity of the marker gene selection process and the need to devise optimal approaches. Future work should aim to expand upon these methods, acknowledging the vast usefulness of marker genes in describing biological systems.

External IDs:doi:10.1007/978-3-662-65902-1_4