DATABASES FOR MOLECULAR MODELING

我们已经开发了用于蛋白质建模的数据库和软件

We have developed databases and software useful for modeling of protein 3-dimensional structure and analysis of sequence-structure relationships. These tools have been distributed freely to biologists and developers of biotechnology software. The work may be divided into three areas: 1) continued development of the "pKB" object-oriented research database, 2) production of "Entrez" sequences for proteins and nucleic acids with known structure, and 3) implementation of a MMDB, a 3-dimensional structure database for "Entrez", PKB has been modified to include core" motif identification, non-redundant set construction, and robust input of protein Data Bank (PDB) files containing new classes of error. It has been used to produce ASN.1-language reports of structural features mappable to sequence, as needed for "Entrez", using the NCBI/GenBank sequence data specification. To include explicit 3-dimensional information in "Entrez" we have created the MMDB database. MMDB employs an ASN.l data specification which defines precisely the chemical graph of biopolymers and non-polymer components such a substrates or cofactors, and maps them into a measured 3-dimensional space. It in this way facilitates generalized homology modeling, inference of spatial structure based on comparison and alignment by chemical structure. MMDB provides data in a form convenient for "Entrez" and for developers of molecular modeling software, as ASN.1 data files and associated object loaders which automatically generate analogous in-memory C data structures. Chemical graphs are constructed largely by computation within pKB, since this information is implicit within the current PDB format, and we have validated the algorithms used. The significance of this work is in providing biologists with easy access to structural data, and in providing a software infrastructure to researchers interested in sequence-structure relationships and programmers developing molecular modeling software.