A databank (3D_-ali) collecting related protein sepquences and structures

An updated version of the 3D_ali databank (Pascarella and\nArgos, 1992) was constructed to incorporate new protein\nstructural and sequence data acquired since the original release\nin 1992. The databank has proved useful in many research\nfields, such as protein sequence and structure analysis and\ncomparison, protein folding, engineering and design, evolution,\nand the like. The collection enhances present protein structural\nknowledge by merging information from proteins having a\nsimilar main-chain fold with homologous primary structures\ntaken from large databases of known sequences. However, the\nconstruction philosophy of the databank has been modified.\nOriginally, the Protein Data Bank (PDB; Bernstein et al,\n1977) of known 3-D structures was exhaustively scanned for\nfold redundancy, and all the possible unique structures were\nincorporated either in multiple structural alignment files or in\nthose containing only one structure where no relatives could\nbe detected. The tertiary structural superpositioning of the\nbackbones, which yielded spatial and topological Co atom\nequivalencing and thus corresponding sequence alignments,\nwas mostly taken from the literature, but was sometimes\ndetermined by the authors using the Rossman-Argos superposition\ntechnique (Argos and Rossmann, 1979). In the updated\n3D_ali databank, only published alignments based on superpositioning\nby the authors of the tertiary structures were\ncollected and only folds with more than one sample structure\nwere considered. Different literature alignments were also\nmerged if they included common folds. As in the former\nrelease, only full coordinate sets with assigned side chains\nwere included, while NMR structures, excluded in the 1992\nrelease, are now incorporated.