Transmembrane protein topology prediction using support vector machines

BMC Bioinformatics

Table 1 SVM per residue accuracy.

SVM	Window size	Kernel	MCC
TM Helix/¬TM Helix	33	RBF	0.80
Inside Loop/Outside Loop	35	Polynomial*	0.63
Re-entrant Helix/¬Re-entrant Helix	27	RBF	0.34
Signal Peptide/¬Signal Peptide	27	RBF	0.76
TM Protein/Globular Protein	33	RBF	0.78

Column 2: Window size – the size of the sliding window in residues. Column 3: Kernel – SVM kernel type. RBF = radial basis function. Column 4: MCC – Matthews correlation coefficient. * The Inside Loop/Outside Loop SVM was trained using a third-order polynomial kernel.

ISSN: 1471-2105