El otro día peleándome con TM-align para conseguir alineaciones de distintos pdbs que parecieran razonables me di cuenta que dicho programa tiene problemas con las proteínas multi-dominio. La solución que le encontré fue en bajarme los PDBs y luego separar las chains. Aquí os dejo el script.

[code language=”python”]

import re import os

#Idea: https://www.biostars.org/p/59715/

list_path = ‘list-DLG4_HUMAN.txt’ db_path = ‘/n/scratch2/rr191/databases/pdb2/’

with open(list_path, ‘r’) as f_pdb: for pdb in f_pdb:

#Getting the info from the pdb to download structure = pdb.split(‘:’)[0].lower() chain = pdb.split(‘:’)[1].upper()[0]

#downloading the pdbs if(not os.path.isfile(db_path+structure+’.pdb’) or True): _ = os.popen(‘wget -O ‘+db_path+structure+’.pdb https://files.rcsb.org/download/’+structure+’.pdb > /dev/null 2>&1’).read()

with open(db_path+structure+’.pdb’, ‘r’) as f: with open(db_path+’chains/’+structure+’_‘+chain+’.pdb’, ‘w’) as fo: for line in f: #selecting the atoms from the desired chain and writing them in a new file if ‘ATOM’ in line and ‘ ‘+chain+’ ‘ in line: fo.write(line)

[/code]