Prediction of essential proteins based on gene expression programming

BMC Genomics

Table 1 Comparison between GEP and the methods using individual feature

Methods	SN	SP	FPR	PPV	NPV	F-measure	ACC	MCC
GEP	0.5467	0.8653	0.1347	0.5467	0.8653	0.5467	0.7927	0.4120
DC	0.4002	0.8217	0.1783	0.4002	0.8217	0.4002	0.7251	0.2219
BC	0.3505	0.8069	0.1931	0.3505	0.8069	0.3505	0.7023	0.1574
CC	0.3548	0.8082	0.1918	0.3548	0.8082	0.3548	0.7043	0.1630
SC	0.3676	0.8120	0.1880	0.3676	0.8120	0.3676	0.7102	0.1796
EC	0.3676	0.8120	0.1880	0.3676	0.8120	0.3676	0.7102	0.1796
IC	0.4010	0.8220	0.1780	0.4010	0.8220	0.4010	0.7255	0.2230
NC	0.4353	0.8321	0.1679	0.4353	0.8321	0.4353	0.7412	0.2674
PeC	0.4036	0.8227	0.1773	0.4036	0.8227	0.4036	0.7267	0.2263
ION	0.5124	0.8551	0.1449	0.5124	0.8551	0.5124	0.7766	0.3675
WDC	0.4576	0.8390	0.1610	0.4580	0.8388	0.4578	0.7516	0.2967

The proteins in PPI network are ranked in descend order according to the scores assigned by our classifier as well as these existing methods. we select top 1167 proteins ranked by each method as candidate essential proteins. The rest of 3926 (= 5093-1167) proteins are regarded as non-essential proteins. According to known essential protein, the values of sensitivity (SN), specificity (SP), positive predictive value (PPV), FPR, negative predictive value (NPV), F-Measure, accuracy (ACC) and Matthews Correlation Coefficent (MCC) are calculated for each method. The table lists the results.

ISSN: 1471-2164