Stochastic algorithm for kinase homology model construction.

摘要: A stochastic algorithm for constructing multiple loops in homology modeling of proteins is presented. The discards variable values iterations based on a cost function and statistical analysis results. Values that remain are used an ensemble best solutions. In test cases, the retains all solutions, compared to exhaustive scan full set conformations. Individual constructed by adding dipeptide units. Dipeptide conformations extracted from database their include bond lengths angles. Single both N- C- terminals center, loop closure evaluated combination penalties peptide Miyazawa-Jernigan (MJ) [1] residue-residue interactions with rest protein. Large ensembles each clustered re-evaluated refined [2] energy term. reduced, then employed construct simultaneously loops. was applied six c-Src kinase family proteins, incorporating total 37-40 residues. RMSD reconstructing 1.45A Lck (structure 1QPE Protein Data Bank) 2.54A human 1FMK). lowest have higher values, 2.06A 3.09A, respectively. average first 1000 3.00A 3.46A, Models “open” structures Jak-2 were basis 1QPE. model found be more flexible region than its c- Src counterpart.