Table 1. Summary of training and testing nsSNVs datasets.

	Name	Source	Number of pathogenic nsSNVs	Number of neutral nsSNVs	Removed variants overlapping with	Notes
Training variants	iFishTrainVar	Pathogenic: HGMD 2014.4	58,794	70,189	—	Common nsSNVs (AF  1%) in 1000 Genomes and dbSNP were used as neutral variants.
		Neutral: 1000 Genomes dbSNP138
Testing variants	NovelVar	Pathogenic: HGMD 2015.3	4,044	4,868	iFishTrainVar	Common nsSNVs (AF  1%) in dbSNP were used as neutral variants.
		Neutral: dbSNP142			PolyPhen2 training variants
	SwissvarFilteredMix	UniProt	1,217	1,324	iFishTrainVar	Only variants on “mix” genes were included.
					PolyPhen2 training variants
	VaribenchSelectedPure	Varibench	2,144	3,777	iFishTrainVar	Only variants on “pure” genes were included.
					PolyPhen2 training variants