Research On The Classification Method Of All-β Protein Fold Type

The Research on protein folding is the frontier of life science, and foldclassification is the foundation of protein in folding study. Nowadays protein foldclassification relies on experts and different database has different classificationstandards, so we urgently need to establish a unified principle method for proteinfolding type classification and classification database.Based on the earlier work, we have built a protein fold type database–LIFCA,which has a uniform principle according to the topological connection and spatialarrangement of secondary structure segments. Based on LIFCA, this paper presents anew scoring function for β-sheet classification. The main work of this paper includes:1. Establishment of the β-sheet template databaseWe select31kinds fold type of β-sheet from LIFCA whose number of samples greaterthan two. Through the analysis of structure features, we determine the templates of31kinds fold type, and extract the corresponding characteristic parameters of thesetemplates from DSSP database. The spatial information mainly are the number of corefragments and the spatial location of residues in the core fragments.2. Establishment and evaluation of PH domain-like barrel folding classificationBased on the classification of SCOP1.75,61proteins (<25％sequence identity) ofPH domain-like barrel are selected as experimental set. Using the contrast informationbetween the template and the spatial structure of the test proteins, this paper presents anew scoring function Fscore. Establishes a protein folding type classification methodbased on Fscore and applies it to this folding type classification. In Astral1.75database, we select16712proteins (<95％sequence identity) as the testing sample,the sensitivity is99.97%. The result indicates that characteristic parameters describethe nature of the folding type, the scoring function Fscore and the classificationmethod based on Fscore can be used for automatic classification of PH domain-likebarrel protein folding type.3. Different checking of β-sheet fold type and software implementationLIFCA database contains574samples of31kinds β-sheet fold type. We give theresult of Sensitivity and Specificity using Fscore and template information of31kindsβ-sheet fold type. Meanwhile, we give the results of the TM-score as a comparison.Though the results of Sensitivity are similar to TM-score results, the results ofSpecificity are better than the TM-score results.The fold type classification mentioned above can be automatically accomplishedby the LIFCA-Fscore program.The β-sheet fold type classification method can not only guide research of otherfold types but also lay the foundation for automatic fold type classification.