Table 2. Machine learning methods and training sets for variant prediction.

Summary of variant prediction methods based on machine learning techniques. Each of these methods utilizes various conservation metrics, training sets, and machine learning techniques for the final prediction.

Name	URL	Sequence Conservation Metric	Machine Learning Method	Variants used for Training
AUTO-MUTE [127]	http://proteins.gmu.edu/automute/	None	SVM and Random Forest	ProTherm
CUPSAT [114]	http://cupsat.tu-bs.de/	None	SVM (Multiple Regression)	ProTherm
I-Mutant2.0 [115]	http://gpcr2.biocomp.unibo.it/~emidio/I-Mutant2.0/I-Mutant2.0_Details.html	None	SVM	ProTherm
LS-SNP [116]	http://modbase.compbio.ucsf.edu/LS-SNP/	SIFT	SVM	dbSNP
MutationTaster [126]	http://www.mutationtaster.org/	Positional conservation in comparison to homologous sequences	Naive Bayes	OMIM, HGMD, common variants from dbSNP
MutPred [122]	http://mutpred.mutdb.org/	Internal, position-specific conservation scorescore	Random Forest	UniProt/Swiss-Prot, HGMD, somatic cancer variantsvariants
nsSNPAnalyzer [123]	http://snpanalyzer.uthsc.edu/	Substitution frequencies in multiple sequence alignments	Random Forest	UniProt/Swiss-Prot
Parepro [117]	http://www.mobioinfor.cn/parepro/	Substitution frequencies of surrounding positions	SVM	HumVar, HumVarProf, and NewHumVar
PhD-SNP [43]	http://snps.biofold.org/phd-snp/phd-snp.html	Substitution frequencies in multiple sequence alignments	SVM	HumVar
PMUT [124]	http://mmb2.pcb.ub.es:8080/PMut/	Substitution frequencies from PSI-BLAST	Feed Forward Neural Network	UniProt/Swiss-Prot
PolyPhen-2 [42]	http://genetics.bwh.harvard.edu/pph2/	PSIC	Naive Bayes	HumDiv and HumVar
SNAP [125]	https://rostlab.org/services/snap/	PSIC	Feed Forward Neural Network	PMD
SNPs&GO [118–120]	http://snps-and-go.biocomp.unibo.it/snps-and-go/	subPSEC	SVM	UniProt/Swiss-Prot
SNPs3D [121]	http://www.snps3d.org/	Substitution frequencies and Shannon entropy	SVM	HGMD, substitutions in homologous sequences