Prediction of protein interaction types based on sequence and network features

BMC Systems Biology

Table 5 Ranking of the top 40 features for the obligate and non-obligate classes based on their Wilcoxon ranked sum test P-values.

Rank	Feature name	Mean obligate	Mean non-obligate	P-value	Rank	Feature name	Mean obligate	Mean non-obligate	P-value
1	MeanSim	0.816	0.63	7.2e-48	21	33	0.0095	0.0148	2e-08
2	CC	0.9	0.748	1.7e-47	22	45	0.0049	0.00841	6.5e-08
3	BP	0.768	0.517	2.7e-41	23	50	0.0040	0.00737	8.6e-08
4	DisRegions	2.17	7.27	9.3e-39	24	57	0.0016	0.00395	2.7e-07
5	FracDisAS	0.0737	0.163	2.4e-35	25	30	0.0224	0.0281	5.3e-07
6	MaxDisLen	17.5	42.2	1.9e-31	26	Degree	35.4	59	7.7e-07
7	MF	0.779	0.626	4.3e-24	27	31	0.015	0.0216	2.1e-06
8	ELM	96.2	151	1.3e-18	28	39	0.0155	0.0193	3.9e-05
9	Betweeness	2463	4315	4.8e-16	29	12	0.0027	0.00132	0.0001
10	18	0.0293	0.04	6.5e-14	30	25	0.0225	0.0148	0.0001
11	26	0.019	0.0274	7.8e-13	31	8	0.0122	0.0082	0.0001
12	17	0.0239	0.0326	7.4e-12	32	52	0.0079	0.00432	0.0001
13	32	0.0109	0.0177	4.2e-11	33	68	0.0017	0.00065	0.0001
14	23	0.0122	0.0181	4.3e-11	34	51	0.0065	0.00398	0.0007
15	7	0.0105	0.0141	2.1e-09	35	41	0.0082	0.00484	0.0008
16	4	0.0173	0.0222	6.3e-09	36	20	0.0298	0.0353	0.001
17	2	0.0043	0.00258	6.9e-09	37	24	0.0216	0.0159	0.0017
18	6	0.0045	0.00752	8e-09	38	44	0.0058	0.00858	0.0022
19	43	0.0037	0.00804	1e-08	39	11	0.0031	0.00161	0.0026
20	5	0.0267	0.0323	1.2e-08	40	49	0.0027	0.00114	0.0028

The numbers in the name column refer to EGDV values for orbits (see Table 4). For the number of disordered binding regions, fraction of disordered amino acids, and ELM both values for protein A and B are combined into one distribution which has two values for each interaction. The Wilcoxon P-value is then calculated for this distribution.

ISSN: 1752-0509