Fig. 4From: An unsupervised deep learning framework for predicting human essential genes from population and functional genomic dataPredictive power of DeepLOF and alternative methods in distinguishing essential genes from putatively nonessential genes. a Performance in predicting ClinGen haploinsufficient genes. b True positive rates in predicting ClinGen haploinsufficient genes at a fixed false negative rate of 5%. Error bars represent bootstrap standard errors of true positive rates. c Performance in predicting human orthologs of mouse essential genes. d Performance in predicting human genes essential for the survival of cell linesBack to article page