Fig. 2From: A flexible network-based imputing-and-fusing approach towards the identification of cell types from single-cell RNA-seq dataOverall procedure of the network-based imputation model in NetImpute framework. The statistic method is used to identify noise data items at both the low-expression and high-expression levels. For each gene’s noise items (star), NetImpute integrates the neighborhood genes in networks (PPI network or gene pathways) that have high-confidence expression values in a specific fuzzy cell subpopulation to train a regression model, and then the learned regression model is used to impute the noise dataBack to article page