Subcellular location prediction of proteins using support vector machines with alignment of block sequences utilizing amino acid composition

Table 3 Prediction accuracy for WoLF PSORT plant data sets. "Chlo," "cyto," "cysk," "E.R.," "extr," "golg," "mito," "nucl," "pero," "plas," "vacu" indicate, chloroplast, cytosol, cytoskeleton, endoplasmic reticulum, extracellular, Golgi apparatus, mitochondria, nuclear, peroxisome, plasma membrane, vacuolar membrane respectively.

Predictor	Location	No. of seq.s	Sensitivity	Specificity	MCC	Average MCC	Overall Accuracy
Our method (2391 sequences)	chlo	750	0.9733	0.8795	0.8893	0.8343	0.8703
	cyto	432	0.9329	0.8258	0.8491
	cysk	41	0.7561	1.0000	0.8677
	E.R.	69	0.7391	0.9623	0.8395
	extr	114	0.8246	0.7581	0.7797
	golg	29	0.6207	0.9474	0.7645
	mito	210	0.7857	0.9066	0.8303
	nucl	456	0.8509	0.8788	0.8329
	pero	52	0.7500	0.9512	0.8417
	plas	165	0.7212	0.9675	0.8255
	vacu	73	0.7671	0.9655	0.8569
Horton et al. (2006) (2426 sequences)							0.86

Note: In our method, although gap penalty = 0.4, γ = 2.5, c = 10, w = 20 are given, posconstraint and negconstraint are not specified.

ISSN: 1471-2105