Predicting cancer-associated germline variations in proteins

BMC Genomics

Table 5 Prediction of the disease type with a SVM-based method

Encoding	Q	MCC	Sp(C)	Sn(C)	Sp(O)	Sn(O)
mut_E_W1	0.59	0.12	0.21	0.55	0.88	0.60
mut_E_W5	0.64	0.17	0.24	0.55	0.88	0.66
OnlyGO	0.89	0.60	0.7	0.63	0.93	0.95
mut_GO_E	0.89	0.60	0.68	0.65	0.93	0.94
mut_GO_E_W1	0.89	0.58	0.68	0.64	0.93	0.94
mut_GO_E_W5	0.89	0.58	0.68	0.64	0.93	0.94
mut_GO	0.90	0.61	0.69	0.66	0.93	0.94

mut= is a 20 elements vector that encodes for the variation type; Wx= a input sequence window of dimension x centered into the variation; E= the evolutionary information on the variation obtained by extracting the 4 columns that represent the wild-type and the mutant residues as reported by PSI-BLAST PSSM/PROFILE output (-Q option); GO= the 3 GO scores. For a definition of classes and scoring indexes see section: Measuring the performance.

ISSN: 1471-2164