def process_ppi1(irefindex_file,
id_logfile,
output_logfile,
biochem_file,
binary_file,
complexes_file,
obo_file,
biogrid_ptm_codes_file,
filtered_pmids_file=None,
accepted_taxids=None):
counts = Phase1Counts()
filtered_pmids = read_filtered_pmids(filtered_pmids_file)
id_map = read_id_mapping(id_logfile)
obo_fp = open(obo_file, 'rU')
ontology = obo.OBOntology(obo_fp)
biochem_filter = _BiochemFilter(ontology, biogrid_ptm_codes_file)
complex_filter = _ComplexFilter(ontology)
input_fp = open(irefindex_file, 'rU')
removed_fp = NullFile()
biochem_fp = open(biochem_file, 'w')
binary_fp = open(binary_file, 'w')
complex_fp = open(complexes_file, 'w')
logfile_fp = open(output_logfile, 'w')
output_fps = (removed_fp, binary_fp, complex_fp, biochem_fp)
for fp in output_fps:
Interaction.write_header(fp)
scanner = parse_mitab_file(input_fp, full_mitab_iterator, None,
iRefIndexInteraction)
for interaction, lines in scanner:
line_numbers = lines[1]
res = _process_interaction(interaction, id_map, filtered_pmids,
logfile_fp, counts, line_numbers, ontology,
biochem_filter, complex_filter,
accepted_taxids)
interaction.to_file(output_fps[res])
counts.to_file(logfile_fp)
input_fp.close()
obo_fp.close()
logfile_fp.close()
for fp in output_fps:
fp.close()
def process_ppi2(input_file,
output_file,
output_logfile,
skipped_pmids_file=None,
max_complex_size=120,
min_complex_size=3):
counts = Phase2Counts()
logfile_fp = open(output_logfile, 'w')
skipped_pmids = read_filtered_pmids(skipped_pmids_file)
skipped_pmids.add(0) # this is invalid pmid - not really a paper
input_fp = open(input_file, 'rU')
output_fp = open(output_file, 'w')
Interaction.write_header(output_fp)
deflator = ComplexDeflator(logfile_fp, max_complex_size, min_complex_size)
for pmid, pairs, complexes in _parse_by_pmid(input_fp):
counts.initial_pairs += len(pairs)
counts.C += len(complexes)
counts.pmids += 1
if pmid in skipped_pmids or len(pairs) < (min_complex_size - 1):
new_complexes = []
unused_pairs = pairs
else:
new_complexes, unused_pairs = deflator(pmid, pairs, complexes)
_write_unused_pairs(output_fp, unused_pairs)
_write_existing_complexes(output_fp, complexes)
_write_new_complexes(output_fp, new_complexes)
counts.unused_pairs += len(unused_pairs)
for intr in new_complexes:
code = intr.edgetype
counter = getattr(counts, code)
counter += 1
setattr(counts, code, counter)
counts.to_file(logfile_fp)
input_fp.close()
output_fp.close()
logfile_fp.close()
