Computational approach to predict species-specific type III secretion system (T3SS) effectors using single and multiple genomes

Table 2 Statistical analysis of attributes chosen by the feature selection methods for the four organisms. Four attributes (PEPIB, CAI, N₃₀ disorder, and G + C content) out of the 21 appeared in at least three of the four organisms tested. Actual values for all attributes are given in Additional file 1)

Organism	Attribute	Positive set		Negative set
Organism	Attribute	Average	SD^b	Average	SD
E. coli	Non-Polar	48.67	8.89	58.42	10.70
	PEPIB	0.86	0.07	0.77	0.14
	CAI	0.58	0.02	0.71	0.06
	N₃₀ disorder	0.53	0.11	0.38	0.13
P. syringae	Tiny	41.11	9.45	30.32	9.09
	Charge^a	0.93	1.25	0.60	2.40
	pI^a	7.74	1.42	6.71	1.76
	PEPIB	0.90	0.11	0.74	0.16
	Aliphatic index	61.19	18.33	100.96	32.15
	CAI	0.51	0.06	0.68	0.07
	G + C content	51.62	5.77	59.04	3.91
	N₃₀ disorder	0.67	0.07	0.41	0.14
S. dysenteriae	Non-Polar	45.11	9.07	56.09	10.11
	Tiny^a	24.22	10.87	29.16	9.29
	G + C content	34.62	1.80	51.65	3.70
	N₃₀ disorder	0.47	0.12	0.22	0.13
S. Typhimurium	PEPIB	0.88	0.12	0.74	0.21
	Instability index	65.21	22.55	38.43	22.54
	CAI	0.58	0.04	0.69	0.05
	G + C content	43.99	6.18	52.18	5.74

^aAttributes that are not statistically different between the positive and negative sets
^b SD standard deviation

ISSN: 1471-2164