Knowledge-based protein modelling

• Main Author: Overington, John Paul
• Published: Birkbeck (University of London) 1991
• Subjects: 572
• Online Access: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.715106

The automated protein modelling program COMPOSER is tested and improved. A test case of the model building of trypsin is described. Prior to the enhancements made to the program the RMSD of the automatically built model was 3.46A, after the improvements this figure is reduced to 1.58A. The program was applied to two ’real-life’ problems from the pharmaceutical industry. The first is the modelling of the serine proteinase domain of tissue-type plasminogen activator. Predictions are made as to residues likely to be important in binding specific endogenous inhibitors. The second example is the modelling of the proteinase from HIV-l on the basis of structures of the distantly related aspartic proteinases and later on the more similar structure of RSV proteinase. The model was later used in the molecular replacement derived solution of the x-ray structure of HIV -1 proteinase.