INTERNET RESOURCES FOR MOLECULAR 3D STRUCTURE

我们已经开发了有用的数据库和软件

We have developed databases and software useful for comparative analysis of protein three-dimensional structure. These tools are distributed freely to biologists and developers of biotechnology software. These software systems fall in three categories, PKB, a database for computational biology researchers, MMDB, a database for multi-platform software developers, and CN3D, an end-user 3D visualization tool for the molecular information browser Entrez. PKB is an object oriented system based on the S language. In the past year PKB has been extended by addition of protein threading routines, and is for this reason now in use in many research laboratories. New structure validation routines have also been added to PKB, and it is for this reason now used routinely at the Protein Data Bank, to identify and correct errors in newly-deposited 3D data files. MMDB is an ASN.1 database where all data items describing macromolecular structure are validated and explicitly listed, so that application software need not contain the complex logic required to retrieve this information from PDB files. Work this year has concentrated on MMDB-API (applications programming interface to MMDB), the in-memory representation of 3D structure information as a C data structure. Useful derived data have also been added, including uniformly defined secondary structure and domain boundary annotation, and complete cross references to NCBI sequence databases and Medline. The MMDB is now distributed as the "structure" database in Entrez, and its API as part of the NCBI toolkit. CN3D (`see in three dimensions'), written this year, is a multi-structure visualization program distributed as part to the Entrez client software, and in a stand-alone version lauchable via the MIME protocol in World-Wide-Web Entrez. The software differs from other public domain viewers in supporting display of multiple aligned structures, from the Entrez "structure neighbors" database, and in supporting simultaneous highlighting/picking of multiple sequence and multiple structure alignments. These features are intended to facilitate molecular biologist's identification of important structure-function relationships within protein families.