Method to group domains

[danielrebib]

Hello,

Thank you for creating a great resource.

Apologies if this is a trivial question, but I am wondering whether there is a method to group multiple domains. I often have hundreds of potential TED annotations for a given protein, and am wondering whether the hits I have are similar/different from one another. Therefore, if the domains themselves are in specific clusters, or belong to specific proteins, having this information would be great.

[shaunmk]

Hi,
I'm not sure I understand your question. When you say you have hundreds of annotations for a given protein, are you referring to TED segmentations and annotations for homologous sequences? Or are you referring to matches of a query domain to TED domains? Or something else entirely?

TED domains that have putative CATH annotations should structurally align (and should correspond to a grouping in CATH), though this will be affected by occasional chopping and classifcation errors.

[danielrebib]

Hi, sorry for the ambiguity. I mean the latter: Matches of a query domain to hundreds of potential TED domains.

I believe your answer suggests that I can take the CATH annotation for the TED domains and see if they are the same/similar? Is there a file that has this information? Thanks!

[shaunmk]

Are you looking for text descriptions of each CATH H- or T-level family? You can get that from this file from the CATH database FTP site (you can download it with wget or similar):

ftp://orengoftp.biochem.ucl.ac.uk/cath/releases/all-releases/v4_3_0/cath-classification-data/cath-names-v4_3_0.txt

Not all nodes in the hierarchy have been assigned names, though.