The MEPI is a web server to predict catalytic residues from protein structures, which integrates the Micro-Environment(ME) and Position Information of each residue.

In our work, we focus on extracting new features from protein structures.

Firstly, with the assumption that catalytic residues have different micro-environment (ME) with non-catalytic ones, a novel feature termed as 'MEscore' is introduced. The ME of a residue is represented as a series of spatially neighboring residue pairs, whose propensities of appearance in the catalytic ME are deduced form a strictly pre-prepared enzyme dataset.

Secondly, Given that catalytic residues prefer to be located in the structural center of protein, we put forward a simple but clear feature ' Dscore' to characterize this tendency.

Thirdly, a novel nonlinear way is used to integrate local property (MEscore) and global property (Dscore) of each residue and the resulting new feature 'MEDscore' is also introduced.

Lastly, since 'MEDscore' shows an excellent complement to conservation score (CONscore), so we introduce a new feature termed as 'CMEDscore', which integrates MEDscore and CONscore with a linear way. CMEDscore may provide more accurate catalytic residue prediction results for users.

For More details:

Lei Han, Yong-Jun Zhang, Jiangning Song, Ming S.Liu and Ziding Zhang. Catalytic Residue Prediction Using a Novel Structural Feature that Integrates the Micro-environment and Geometrical Location Properties of Residues. PLos ONE 7(7):e41370.