ESLpred2: improved method for predicting subcellular localization of eukaryotic proteins

Garg, Aarti; Raghava, Gajendra PS

doi:10.1186/1471-2105-9-503

BMC Bioinformatics

Table 2 The detailed prediction results of different modules and comparison of performance with BaCelLo method on non-redundant and organism specific datasets

From: ESLpred2: improved method for predicting subcellular localization of eukaryotic proteins

Datasets	Localizations	PSI-BLAST (A)		(PSSM+AAC-NTerm) (B)		Hybrid2 (A+B)		Hybrid2 (10-fold CV)		^#Using BaCelLo strategy (B)	BaCelLo ¹⁸ method
		ACC	MCC	ACC	MCC	ACC	MCC	ACC	MCC	ACC	ACC (Third level)
Fungi dataset	Cytoplasm	10.9	----	53.6	0.32	54.0	0.36	51.7	0.37	62.6	60.2
	Mitochondria	12.2	----	84.0	0.75	82.5	0.77	83.5	0.77	90.4	81.4
	Nuclear	39.7	----	73.0	0.73	78.6	0.59	80.7	0.60	74.7	67.1
	Extracellular	29.6	----	92.1	0.92	92.1	0.93	93.2	0.93	94.3	94.3
	*Overall*	29.5	----	*72.7	0.56	75.9	0.60	77.0	0.61	80.5	70.1
	*Average*	23.1	----	*75.7	0.63	76.8	0.66	77.3	0.67	76.5	75.8
		ACC	MCC	ACC	MCC	ACC	MCC	ACC	MCC	ACC	*ACC (Third level)*
Animal dataset	Cytoplasm	28.7	----	62.9	0.42	63.3	0.49	61.3	0.48	70.6	65.3
	Mitochondria	17.0	----	77.1	0.75	78.2	0.77	78.7	0.77	91.5	76.1
	Nuclear	53.8	----	69.0	0.60	77.7	0.68	79.1	0.69	72.6	64.8
	Extracellular	40.9	----	92.4	0.86	95.3	0.90	95.0	0.90	93.8	90.8
	*Overall*	42.9	----	*75.8	0.66	80.8	0.72	81.0	0.73	80.1	73.8
	*Average*	35.0	----	*75.4	0.66	78.6	0.71	78.5	0.71	82.1	74.2
		ACC	MCC	ACC	MCC	ACC	MCC	ACC	MCC	ACC	*ACC (Fourth level)*
Plant dataset	Chloroplast	31.4	----	77.5	0.67	81.9	0.69	82.8	0.71	90.7	73.0
	Cytoplasm	6.90	----	51.7	0.50	50.0	0.53	50.0	0.50	79.3	51.7
	Mitochondria	16.4	----	67.2	0.66	65.8	0.63	70.2	0.66	67.2	50.7
	Nuclear	48.8	----	80.2	0.77	81.8	0.76	81.8	0.79	86.8	71.9
	Extracellular	26.8	----	87.8	0.65	90.2	0.70	95.1	0.76	85.4	85.4
	*Overall*	30.3	----	*74.5	0.66	76.6	0.68	78.0	0.70	84.7	68.2
	*Average*	26.4	----	*72.9	0.64	73.9	0.67	76.0	0.69	81.9	66.6

ACC is accuracy; MCC is Matthew correlation coefficient; ACC is calculated in percentage
*Overall and average accuracy obtained at SVM parameters: For Fungi dataset (kernel = RBF, γ = 5, C = 4); Animal dataset (kernel = RBF, γ = 5, C = 2); Plant dataset (RBF, γ = 9, C = 3).
^# SVM parameters obtained for each class using hybrid1 features-For Fungi dataset (Cytoplasm: j = 4, γ = 7, C = 0.4, threshold value = 0.0; Mitochondria: j = 5, γ = 1, C = 1.6, threshold value = 0.0; Nuclear: j = 4, γ = 7, C = 0.54, threshold value = 0.0; Extracellular: j = 3, γ = 1, C = 1, threshold value = 0.0), Animal dataset (Cytoplasm: j = 3, γ = 9, C = 0.5, threshold value = 0.0; Mitochondria: j = 25, γ = 1, C = 2, threshold value = 0.0; Nuclear: j = 3, γ = 9, C = 0.5, threshold value = 0.0; Extracellular: j = 6, γ = 2, C = 1, threshold value = -0.1), Plant dataset (Cytoplasm: j = 2, γ = 3, C = 0.7, threshold value = 0.1; Mitochondria: j = 1, γ = 5, C = 75, threshold value = 0.2; Nuclear: j = 2, γ = 3, C = 0.7, threshold value = 0.1; Extracellular: j = 9, γ = 1, C = 1, threshold value = 0.0)

Back to article page

ISSN: 1471-2105

Contact us

General enquiries: journalsubmissions@springernature.com