Fig. 3

A The AUROC values of SRGS and each compared GRN method tested on simulated single-cell datasets without dropouts. The datasets contain two gene sizes (50 and 100) and six sample sizes (10, 50, 100, 5,001,000 and 5000). For each dataset, five gold-standard networks (Ecoli1, Ecoli2, Yeast1, Yeast2, and Yeast3) were used as references. B The mean and standard deviation of AUROC of each GRN method across all gene sizes, sample sizes and networks shown in (A). C The AUROCs of SRGS and each compared GRN method tested on simulated single-cell datasets with dropouts. The datasets contain two gene sizes (50 and 100), two sample sizes (50 and 100), and three dropout rates (20, 50, and 70%). For each dataset, five gold-standard networks (Ecoli1, Ecoli2, Yeast1, Yeast2, and Yeast3) were used as references. D The mean and standard deviation of AUROC of each GRN method across all gene sizes, sample sizes, networks, and dropout rates shown in (C). The benchmarking methods are GENIE3, SINCERITIES, ARACNE, CLR, MRNET, PIDC, PPCOR, LEAP, BoolNet, and bnlearn. For SINCERITIES, the datasets with sample sizes of 10 and 50 could not be tested