How Can I Get All The Domain Structures In A Cath Superfamily?

Question

[Simon Cockell]

I want to retrieve all of the PDB files for domains that belong to a particular superfamily in the CATH database.

I can get representative structures of families from the CATH download area, I can also get lists of domains and domain boundaries, so I can construct them myself by pulling files from the PDB and trimming programatically, but I was wondering if anyone here has come up with a more efficient method?

I'll proceed with writing the code and post it as an answer once I'm done, but maybe someone can provide me with a better solution in the meantime.

[Andrew Clegg]

Hi, Andy from CATH here :-)

Do you mean the original whole-protein PDBs from which the domains were extracted, or the chopped PDBs we produce that just contain individual domains?

Answer 1

[Andrew Clegg]

Sorry, ignore my (deleted) comment, I read your question again and saw the bit about trimming.

Once you have the domain list, you can hit this URL to get an individual domain:

http://www.cathdb.info/api/data/pdb/1cukA01

Where 1cukA01 is the domain in question. Or for a whole chain:

http://www.cathdb.info/api/data/pdb/1cukA

... or indeed a whole pdb the same way.

We're adding this to the docs at the moment, didn't realise it wasn't in there :-)

Also planning to add a REST service to get all the members of a superfamily via HTTP, but you can parse this out of the domain list as you spotted.

[Neilfws]

It's a couple of years since I used CATH. I'm amazed and impressed by its transformation.

[Simon Cockell]

that's perfect! I love BioStar...

[Andrew Clegg]

Thanks Neil -- spread the love :-)

[Andrew Clegg]

PS... You should see the next version (end of this year)

Answer 2

[Simon Cockell]

The code I used for the answer to this question is can be found here.

	import urllib
	import os, os.path
	from optparse import OptionParser
	def main(superfamily):
	#fetch the list of domains in the superfamily from the CathDomainList
	dom_lst = get_domain_list(superfamily)
	#for each domain, retrieve the PDB file from CATH
	get_domain_structures(dom_lst, superfamily)
	def get_domain_list(superfamily):
	count = 0
	domain_list = []
	sf_tokens = superfamily.split('.')
	fh = open(os.path.join("data", "cath", "CathDomainList"), 'r')
	for line in fh.readlines():
	if not line.startswith('#'): #exclude comment lines
	tokens = line.rstrip().split()
	#if C, A, T and H match, the domain is a member of the right superfamily
	if int(tokens[1]) == int(sf_tokens[0]) \
	and int(tokens[2]) == int(sf_tokens[1]) \
	and int(tokens[3]) == int(sf_tokens[2]) \
	and int(tokens[4]) == int(sf_tokens[3]):
	domain_list.append(tokens[0])
	count += 1
	print("There are "+str(count)+" domains in superfamily "+superfamily)
	return domain_list
	def get_domain_structures(domain_list, superfamily):
	for domain in domain_list:
	#can also get chain and full pdb entries by modifying the URL
	url = 'http://www.cathdb.info/api/data/pdb/'+domain
	pdb = urllib.urlopen(url).read()
	if not os.path.exists(os.path.join('data', 'pdb', superfamily)):
	os.mkdir(os.path.join('data', 'pdb', superfamily))
	out = open(os.path.join('data', 'pdb', superfamily, domain+'.pdb'), 'w')
	out.write(pdb)
	return url
	if __name__ == '__main__':
	parser = OptionParser(usage="Usage: %prog [options]", version="%prog 0.1")
	parser.add_option("-c", dest="c", help="Class", metavar="CLASS")
	parser.add_option("-a", dest="a", help="Architecture", metavar="ARCHITECTURE")
	parser.add_option("-t", dest="t", help="Topology", metavar="TOPOLOGY")
	parser.add_option("-s", dest="s", help="Homologous Superfamily", metavar="SUPERFAMILY") #would have used -h, but it's reserved for help
	(options, args) = parser.parse_args()
	sf = options.c+'.'+options.a+'.'+options.t+'.'+options.s
	main(sf)

view raw get_cath_domains.py hosted with ❤ by GitHub

Much simpler code than I thought would be necessary, thanks to Andrew :)

Answer 3

[Jmdana]

This answer arrives very late but it could help others.

PDBe has a service called PDBeXplore. This service cross-references data between different databases/classification systems and PDBe (one of those classification systems is CATH).

If you go to the PDBeXplore - CATH module and select a CATH class you will be able to see all the PDB entries belonging to it. You will also be able to download the whole table as a tab delimited file (right hand panel). From there, it is very easy just to parse the file and download the PDB files one by one.

Hope that helps.