Identifying digenic disease genes via machine learning in the Undiagnosed Diseases Network

An accurate prediction model of digenic interaction for estimating pathogenic gene pairs of human diseases

“…● DIEP made more accurate predictions on the digenic effects of genetic diseases . Significantly, we had manually curated 64 probably digenic gene pairs from various research studies, in which 14 pairs also served as the positive test set by another predictor for identifying digenic disease genes (DiGePred) [19] . As a result, our DIEP correctly predicted all 14 digenic pairs (100%) with relatively high probabilities (0.54–0.97), while DiGePred only had the TP rate of 57.14% at a suggested threshold of 0.496, and the predicted score only ranged from 0.528 to 0.804.…”

“…Since the DIEP and DiGePred [19] are relatively similar, we compared them systematically and comprehensively from three aspects. First, the PR AUCs (Area Under Precision-Recall Curve) were compared between DIEP and DiGePred based on two different test sets (unaffected-no-gene-overlap_non_digenic_pairs_held_out_test.csv and random-no-gene-overlap_non_digenic_pairs_held_out_test.csv) collected from DiGePred publication ( https://github.com/CapraLab/DiGePred ).…”

“…Besides, Papadimitriou et al developed a variant combinations pathogenicity predictor (VarCoPP [18] ) mainly for predicting disease-causing variant combinations in gene pairs. Mukherjee et al constructed the DiGePred [19] aimed to identify digenic disease-causing gene pairs, which addressed the same problem as our DIEP, but with limitations on the selection of negative training sets and gene-level features for model training.…”

New Developments and Possibilities in Reanalysis and Reinterpretation of Whole Exome Sequencing Datasets for Unsolved Rare Diseases Using Machine Learning Approaches

“…DiGePred, a random forest classifier, has been developed to specifically identify candidate disease gene pairs (digenic diseases) by features derived from biological networks, genomics, evolutionary history and functional annotations [ 65 ]. DiGePred used an ML strategy called ensemble method which has been used in RD classification and is based on random forest classifiers where multiple weak decision trees are combined to generate a better predictive outcome in terms of classification [ 20 ].…”

“…DiGePred used an ML strategy called ensemble method which has been used in RD classification and is based on random forest classifiers where multiple weak decision trees are combined to generate a better predictive outcome in terms of classification [ 20 ]. The use of DiGePred has helped in the discovery of genetic causes for rare non-monogenic diseases by providing a score to evaluate variant gene pairs for the potential to cause digenic disease [ 65 ]. This type of analysis could be then used to assess the prevalence of putative gene pairs in undiagnosed rare non-monogenic diseases.…”