def _get_aln_diff(self, blast_result_files):
file_handlers = FileHandlers()
aln_diff = []
for result_file in blast_result_files:
data = self._open_results_file(result_file)
cleaned_data = []
for i in range(len(data)):
fields = data[i].split(' ')
cleaned = file_handlers.clean(fields)
while cleaned.count('') > 0:
cleaned.remove('')
cleaned_data.append(cleaned)
#for line in cleaned_data:
gene_start = ''
pdb_start = ''
while gene_start == '':
for line in cleaned_data:
if ( len(line) == 4 and line[0] == 'Query' ):
gene_start = int(line[1])
break
else:
pass
while pdb_start == '':
for line in cleaned_data:
if ( len(line) == 4 and line[0] == 'Sbjct' ):
pdb_start = int(line[1])
break
else:
pass
seq_name = result_file.split('_')[0] + '_' + result_file.split('_')[3] + '_' + result_file.split('_')[4]
start_site_difference = gene_start - pdb_start
aln_diff.append((seq_name, start_site_difference))
#print aln_diff
return aln_diff
