Source Code for Module csb.bio.io.procheck

 1  """ 
 2  Procheck parser 
 3  """ 
 4  import os 
 5  import re 
 6  import shutil 
 7   
 8  from csb.io import Shell 
 9  from csb.io import TempFolder 
10   
11 -class ProcheckParser(): 
12      """ 
13      Simple Prochceck Summary parser 
14      """ 
15 -    def __init__(self): 
16          self.binary = 'procheck.scr' 
17          self.acc = 2.0 
18           
19 -    def parse(self, fn): 
20          """ 
21          @param fn: source  file to parse 
22          @type fn: str 
23   
24          @return: dicttionary of parsed quality indicatiors 
25          """ 
26          info = dict() 
27           
28          f_handler = open(os.path.expanduser(fn)) 
29          text = f_handler.read() 
30           
31          input_file_name = re.compile('>>>-----.*?\n.*?\n\s*\|\s*(\S+)\s+') 
32          residues = re.compile('(\d+)\s*residues\s\|') 
33          ramachandran_plot = re.compile('Ramachandran\splot:\s*(\d+\.\d+)' +  
34                                        '%\s*core\s*(\d+\.\d+)%\s*allow\s*(\d+\.\d+)' +  
35                                        '%\s*gener\s*(\d+\.\d+)%\s*disall') 
36          labelled_all = re.compile('Ramachandrans:\s*(\d+)\s*.*?out\sof\s*(\d+)') 
37          labelled_chi = re.compile('Chi1-chi2\splots:\s*(\d+)\s*.*?out\sof\s*(\d+)') 
38          bad_contacts = re.compile('Bad\scontacts:\s*(\d+)') 
39          g_factors = re.compile('G-factors\s*Dihedrals:\s*([0-9-+.]+)' +  
40                                '\s*Covalent:\s*([0-9-+.]+)\s*Overall:\s*([0-9-+.]+)') 
41   
42          info['input_file'] = input_file_name.search(text).groups()[0] 
43          info['#residues'] = int(residues.search(text).groups()[0]) 
44          info['rama_core'], info['rama_allow'], info['rama_gener'], info['rama_disall'] = \ 
45                             [float(g) for g in ramachandran_plot.search(text).groups()] 
46          info['g_dihedrals'], info['g_bond'], info['g_overall'] = \ 
47                               [float(g) for g in g_factors.search(text).groups()] 
48          info['badContacts'] = int(bad_contacts.search(text).groups()[0]) 
49          info['labelledAll'] = float(labelled_all.search(text).groups()[0]) / \ 
50                                float(labelled_all.search(text).groups()[1]) 
51          info['labelledChi'] = float(labelled_chi.search(text).groups()[0]) / \ 
52                                float(labelled_chi.search(text).groups()[0]) 
53   
54          f_handler.close() 
55           
56          return info 
57   
58   
59 -    def run(self, pdb_file): 
60          """ 
61          Runs procheck for the given pdbfile and parses the output. 
62          Will fail if the procheck binary is not in the path. 
63           
64          @param pdb_file: file to parse 
65          @return: dict of parsed values 
66          """ 
67          wd = os.getcwd() 
68          base = os.path.basename(pdb_file) 
69   
70          with TempFolder() as tmp: 
71              shutil.copy(os.path.expanduser(pdb_file), tmp.name) 
72              os.chdir(tmp.name) 
73              Shell.run('{0} {1} {2}'.format(self.binary, 
74                                             os.path.join(tmp.name, base), 
75                                             self.acc)) 
76              summary = '.'.join([os.path.splitext(base)[0], 'sum']) 
77              out = self.parse(os.path.join(tmp.name, summary)) 
78              os.chdir(wd) 
79   
80          return out 
81