Table 1. The predictive performance of our method in a series of cross-validation experiments.

Random MicroRNAs	Leave-One-Out	50a	100a	150a	200a	250a	300a	All microRNAs1
		84.12%c	85.00%c	85.61%c	86.10%c	86.56%c	86.76%c	86.93%
	Ab Initio	50a	100a	150a	200a	250a	300a	All microRNAs1
		69.23%c	68.89%c	69.68%c	70.00%c	70.44%c	71.30%c	70.92%
Random Diseases	Leave-One-Out	90b	100b	110b	120b	130b	140b	All diseases2
		81.53%c	80.66%c	79.44%c	83.43%c	81.52%c	83.47%c	83.50%
	Ab Initio	90b	100b	110b	120b	130b	140b	All diseases2
		77.97%c	77.28%c	79.38%c	79.27%c	79.63%c	79.02%c	80.03%

¹The candidate microRNAs were obtained after deleting the microRNA (microRNAs) related to query disease.

²The candidate diseases were obtained after deleting the disease (diseases) associated with the query microRNA.

^aThe number of randomly selected candidate microRNAs in microRNA prioritization.

^bThe number of randomly selected candidate diseases in disease prioritization.

^cThe AUC score of ROC curve.