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

Table 4 Comparison of predictive accuracy for non-plant proteins in the TargetP data set. "mTP", "SP", and "other" indicate proteins destined for mitochondria, secretory pathway, and other locations (nucleus and cytosol), respectively.

Predictor	Location	Sensitivity	Specificity	MCC	Average MCC	Overall Accuracy
Our method	mTP	0.7647	0.8990	0.8005	0.8452	0.9204
	SP	0.9157	0.9056	0.8790
	other	0.9576	0.9324	0.8560
Matsuda et al. (2005)	mTP	0.8303	0.8635	0.8228	0.8542	0.9229
	SP	0.9091	0.9118	0.8788
	other	0.9498	0.9409	0.8609
Kim et al. (2004)	mTP	0.6483	0.8569	0.7121	0.7635	0.8762
	SP	0.8530	0.8736	0.8158
	other	0.9389	0.8819	0.7626
Emanuelsson et al. (2000)	mTP	0.89	0.67	0.73	0.8233	0.900
	SP	0.96	0.92	0.92
	other	0.88	0.97	0.82
Reczko et al. (2004)	mTP	0.78	0.82	0.77	0.8333	0.913
	SP	0.93	0.91	0.89
	other	0.93	0.94	0.84

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