Fig. 2.

Typical strategies and representative methods for annotating cell subpopulations identified by scRNA-seq. In addition to manual annotation, which is potentially time-consuming and subjective, automated cell type annotation can be mainly sorted into three categories: marker gene-based, reference transcriptome-based, and supervised machine learning-based approaches. The example approach names are listed in the plot. scRNA-seq single-cell RNA sequencing, scCATCH single-cell cluster-based automatic annotation toolkit for cellular heterogeneity, SCINA semi-supervised category identification and assignment, CHETAH characterization of cell types aided by hierarchical classification, SingleR single cell recognition, OnClass ontology-based single cell classification, ACTINN automated cell type identification using neural networks