ANALYSIS OF 6 PROTEIN STRUCTURES PREDICTED BY COMPARATIVE MODELING TECHNIQUES

The protein structures of six comparative modeling targets were predic ted in a procedure that relied on improved energy minimization, withou t empirical rules, to position all new atoms, The structures of human nucleoside diphosphate kinase NM23-H2, HPr from Mycoplasma capricolum, 2Fe-2S ferredoxin from Haloarcula marismortui, eosinophil-derived neu rotoxin (EDN), mouse cellular retinoic acid protein I (CRABP1), and P4 50eryf were predicted with root mean square deviations on C alpha atom s of 0.69, 0.73, 1.11, 1.48, 1.69, and 1.73 Angstrom, respectively, co mpared to the target crystal structures. These differences increased a s the sequence similarity between the target and parent proteins decre ased from about 60 to 20% identity, More residues were predicted than form the common region shared by the two crystal structures. In most c ases insertions or deletions between the target and the related protei n of known structure were not correctly positioned. One two residue in sertion in CRABP1 was predicted in the correct conformation, while a n ine residue insertion in EDN was predicted in the correct spatial regi on, although not in the correct conformation, The positions of common cofactors and their binding sites were predicted correctly, even when overall sequence similarity was low. (C) 1995 Wiley-Liss, Inc.