Fig. 1

Schematic illustration of the scRNA-seq data-based cell type identification framework. (I) Identification of dropout clusters. First, the dropout information was extracted from the gene expression profile, and the dropout feature matrix was constructed. Then, the dropout distances between cells were calculated, and DBSCAN was performed to identify dropout clusters of cells. (II) Construction of the differential co-expression network. We performed differential analysis of the dropout rates and molecule expression levels and constructed a differential co-expression network based on the correlations between differential molecules. (III) Identification of cell types and candidate cell markers. We performed MCL on the differential co-expression network to identify differential modules. We then used these differential modules as new features to identify the cell types and further identified candidate cell markers for each cell type