next up previous
Next: Defining domains Up: New and old data Previous: Adding new chains

Tracking domains

In 3Dee each domain is tagged by a unique identifier. This domain identifier consists of the PDB code, chain identifier, domain number, definition source and revision number. For example 1bia-2-DOMAK refers to the second domain of the protein 1bia; the domain definition is derived from the DOMAK program [Siddiqui and Barton, 1995]. 2fb4h-2-AS.3 refers to the second domain of the `h' chain of protein 2fb4; it is the third revision of the domain definition produced by AS. Other sources for domain definitions in the database currently are GJB, JAC, UDE and AUTO. Domain definitions containing AUTO are produced automatically by sequence alignment with a chain from the same sequence family for which domains have been defined already. By giving each definition a unique source, alternative domain definitions for the same protein may co-exist. For example definitions produced by different available computer methods could be stored. If more than one domain definition for a chain exists, one is chosen as the default definition and used to produce the domain sequence and domain structure families. Classification of domains based on a different criterion, like functional units, may be required. For many proteins in 3Dee, the definitions would be correct as they stand, while some would need different domain definitions. So the serine proteases and aspartic proteases would have to be defined as single domain proteins rather than two domain proteins as they are in the current version of 3Dee. Accordingly, one might wish to use the default definitions for most proteins and extract another domain definition for others. Placing a keyword in the domain definition file allows the correct definition(s) to be selected. Hence, permitting a set of domain definitions based on alternative criteria to be extracted from the same database. A useful consequence of identifying domains in this manner is that domain data, for example structural comparisons between domains performed with the STAMP program [Russell and Barton, 1992], do not have to be regenerated.
next up previous
Next: Defining domains Up: New and old data Previous: Adding new chains
Uwe Dengler, 2000-10-16