def _parse_by_pmid(input_fp):
# Will go smallest to largest pubmed id, in two passes
# First pass: collect pubmed ids with pointers to their interactions
scanner = parse_mitab_file(input_fp, full_mitab_iterator)
pmids_map = offsets_by_pmid_consumer(scanner)
sorted_pmids = sorted((len(v), k) for k, v in pmids_map.iteritems())
# Second pass
for _, pmid in sorted_pmids:
scanner = parse_mitab_file(input_fp, partial_mitab_iterator,
(pmids_map[pmid], ))
pairs, complexes = full_interaction_consumer(scanner)
yield pmid, pairs, complexes
def process_biochem(biochem_file, output_fp, ontology, logfile_fp, counts):
"""
Process all candidate 'biochemical' interactions in biochem_file and write
consolidated interactions to output_fp.
"""
input_fp = open(biochem_file, 'rU')
scanner = parse_mitab_file(input_fp, full_mitab_iterator)
pairs, complexes = full_interaction_consumer(scanner)
counts.pmids.update(intr.pmid for intr in pairs)
counts.pmids.update(intr.pmid for intr in complexes)
# Extract additional interactions from 'complexes', which are in fact
# biochemical reactions described in more detail
pairs2, complexes2 = _extract_biochem_from_complexes(complexes)
# Collect source interactions by (directed) pair of proteins plus pubmed id
pmid_links, unassigned = _collect_by_pmid(pairs + pairs2, _get_biochem_key)
# Find conflicts and inconsistencies
_get_biochem_key_conflicts(pmid_links, logfile_fp)
# Attempt to assign interactions from DIP, which are not directional
unresolved = _collect_biochem_unassigned(unassigned, pmid_links)
# Consolidate each interaction and write to file
sorted_keys = sorted(pmid_links.keys())
for key in sorted_keys:
interactions = pmid_links[key]
counts.biochem += len(interactions)
_correct_biogrid_biochem(interactions)
for new_intr in _consolidate_links(interactions, ontology, 'D',
logfile_fp, counts):
new_intr.to_file(output_fp)
counts.directed += 1
# Write all 'unused' interactions
for item in unresolved:
key = _get_undirected_key(item)
for new_intr in _consolidate_links([item], ontology, 'B', logfile_fp,
counts):
new_intr.to_file(output_fp)
counts.undirected += 1
# Write unused complexes
for cmplx in complexes2:
key = _get_undirected_key(cmplx)
for new_intr in _consolidate_links([cmplx], ontology, 'C', logfile_fp,
counts):
new_intr.to_file(output_fp)
counts.complexes += 1
input_fp.close()
def process_complexes(complexes_file, output_fp, ontology, logfile_fp, counts):
"""
Process all candidate 'complexes' in complexes_file and write
consolidated interactions to output_fp.
"""
input_fp = open(complexes_file, 'rU')
scanner = parse_mitab_file(input_fp, full_mitab_iterator)
offsets_map = _consume_complex_offsets_only(scanner)
# Consolidate each interaction and write to file
sorted_keys = sorted(offsets_map.iterkeys())
for key in sorted_keys:
scanner = parse_mitab_file(input_fp, partial_mitab_iterator,
(offsets_map[key], ))
_, complexes = full_interaction_consumer(scanner)
counts.pmids.update(intr.pmid for intr in complexes)
for new_intr in _consolidate_complexes(complexes, ontology,
logfile_fp):
new_intr.to_file(output_fp)
counts.complexes += 1
input_fp.close()
def __call__(self, ppiTrim_file, *dbargs, **dbkwargs):
self.connect_to_db(*dbargs, **dbkwargs)
input_fp = open(ppiTrim_file, 'rU')
scanner = parse_mitab_file(input_fp, full_mitab_iterator)
for intr, _ in scanner:
self._insert_interaction(intr)
self._insert_all_conflicts()
input_fp.close()
self.cur.close()
self.conn.commit()
self.conn.close()
def process_undirected(binary_file, output_fp, ontology, logfile_fp, counts):
"""
Process all 'other' interactions - undirected physical interactions. Write
consolidated interactions to output_fp.
"""
input_fp = open(binary_file, 'rU')
scanner = parse_mitab_file(input_fp, full_mitab_iterator)
offsets_map = _consume_undirected_offsets_only(scanner)
# Consolidate each interaction and write to file
sorted_keys = sorted(offsets_map.iterkeys())
for key in sorted_keys:
scanner = parse_mitab_file(input_fp, partial_mitab_iterator,
(offsets_map[key], ))
pairs, _ = full_interaction_consumer(scanner)
counts.pmids.update(intr.pmid for intr in pairs)
counts.other += len(pairs)
for new_intr in _consolidate_links(pairs, ontology, 'X', logfile_fp,
counts):
new_intr.to_file(output_fp)
counts.undirected += 1
input_fp.close()
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 summarize_complexes(ppiTrim_file, output_file):
"""
Output a summary of generated complexes of 'A' and 'N' type.
These are randomly shuffled for sampling.
"""
lines = []
input_fp = open(ppiTrim_file, 'rU')
output_fp = open(output_file, 'w')
scanner = parse_mitab_file(input_fp, full_mitab_iterator)
for intr, _ in scanner:
edgetype = intr.edgetype
n = len(intr.interactors)
if not intr.is_complex() or n > 20:
continue
if ('C' in edgetype or 'G' in edgetype or 'R' in edgetype):
continue
if not ('A' in edgetype):
continue
geneids = ', '.join(p.uid.acc for p in intr.interactors)
symbols = ', '.join(p.alias.ids[0].acc for p in intr.interactors)
for item in intr.confidence.ids:
if item.db == 'sourcedbs':
sourcedbs = item.acc
break
line = '\t'.join([
intr.complex.uid.acc,
'%d' % len(intr.interactors), edgetype, sourcedbs,
'%d' % intr.pmid, symbols, geneids, '*****'
])
lines.append(line)
random.shuffle(lines)
for line in lines:
output_fp.write(line)
output_fp.write('\n')
input_fp.close()
output_fp.close()
def extract_protein_ids(mitab_file, filtered_pmids, accepted_taxids):
"""
Extract protein IDs from iRefIndex PSI-MI TAB 2.6 file
"""
protein_ids = set()
nullfile = NullFile()
input_fp = open(mitab_file, 'rU')
scanner = parse_mitab_file(input_fp, full_mitab_iterator, None,
iRefIndexInteraction)
for interaction, lines in scanner:
line_numbers = lines[1]
if is_filtered(interaction, filtered_pmids, nullfile, line_numbers,
accepted_taxids):
continue
for p in interaction.interactors:
# p stands for 'protein interactor'
protein_ids.add(p.id)
input_fp.close()
return protein_ids