Table 2. Disease candidate gene prediction.

	Alzheimer's	Schizophrenia	Parkinson's	Autism	OMIM
Optimum List from GO (ROC AUC)	0.78	0.73	0.73	0.71	0.76+/−0.22
Machine Learning (ROC AUC)	0.74	0.69	0.71	0.67	–

Table 2: We predicted disease genes using the optimal gene ranking based on GO and then validated using disease gene candidate sets (top). These ROC results compare favorably to predicting candidate genes using a strongly performing protein interaction network and a powerful gene function prediction algorithm (bottom). OMIM disease gene sets are too small for cross-validation using a gene function prediction algorithm.