PDB-REPRDB: a database of representative protein chains in PDB (Protein Data Bank)

We propose a novel set of 'representative' protein chains in PDB, where not only sequential but also structural similarities are taken into account. Hobohm et al. have already proposed "PDB_SELECT", which eliminates redundant chains based solely on sequence similarity. "PDB_SELECT" is frequently updated and the latest version is available at EMBL WWW server. In our set of entries "PDB-REPRDB," however, structural similarities are also considered, in order not to overlook local conformation diversity within a group of sequentially similar chains. Our set guarantees that every representative is the best among that similar protein group, regarding experimental or structure-determination quality (i.e. resolution and R-value). The first version (based on PDB Release 70) of PDB-REPRDB was released in 1995 and the second version (PDB Release 78) will be available by April 1997.