def group_go_by_ontology(data, dag, fill_na, accession_to_feature_file):
if isinstance(data, str):
feature_df = load_data_frame(accession_to_feature_file, fill_na=fill_na)
p_go = get_feature_for_accession(feature_df, data, 'uniprot', 'go')
else:
p_go = [t.upper() for t in data if 'go' in t.lower()]
# Separate the namespaces in the go terms.
p_go_cc = filter(lambda x: ontology.id_to_node(x, dag).namespace == 'cellular_component', p_go)
p_go_bp = filter(lambda x: ontology.id_to_node(x, dag).namespace == 'biological_process', p_go)
p_go_mf = filter(lambda x: ontology.id_to_node(x, dag).namespace == 'molecular_function', p_go)
assert len(set(p_go_cc) & set(p_go_bp) & set(p_go_mf)) == 0
return {'cc': p_go_cc, 'bp': p_go_bp, 'mf': p_go_mf}
def compute_ss(ppi_tuples):
r_file_in = tempfile.mktemp(suffix='.tsv', prefix='r_in_', dir='tmp')
r_file_out = tempfile.mktemp(suffix='.tsv', prefix='r_out_', dir='tmp')
dag = ontology.load_go_dag(OBO_FILE)
feature_df = load_data_frame(ACCESSION_FEATURES_FILE, fill_na=np.NaN)
# Write the three seperate GO columns to the r_input_file
fp = open(r_file_in, 'w')
fp.write("p1\tp2\tp1_go_cc\tp2_go_cc\tp1_go_bp\tp2_go_bp\tp1_go_mf\tp2_go_mf\n")
for p1, p2 in ppi_tuples:
p1_go = get_feature_for_accession(feature_df, p1, 'uniprot', 'go')
p2_go = get_feature_for_accession(feature_df, p2, 'uniprot', 'go')
# Separate the namespaces in the go terms.
p1_go_cc = set(filter(lambda x: ontology.id_to_node(x, dag).namespace == 'cellular_component', p1_go))
for p in p1_go_cc:
assert ontology.id_to_node(p, dag).namespace == 'cellular_component'
p2_go_cc = set(filter(lambda x: ontology.id_to_node(x, dag).namespace == 'cellular_component', p2_go))
for p in p2_go_cc:
assert ontology.id_to_node(p, dag).namespace == 'cellular_component'
p1_go_bp = set(filter(lambda x: ontology.id_to_node(x, dag).namespace == 'biological_process', p1_go))
for p in p1_go_bp:
assert ontology.id_to_node(p, dag).namespace == 'biological_process'
p2_go_bp = set(filter(lambda x: ontology.id_to_node(x, dag).namespace == 'biological_process', p2_go))
for p in p2_go_bp:
assert ontology.id_to_node(p, dag).namespace == 'biological_process'
p1_go_mf = set(filter(lambda x: ontology.id_to_node(x, dag).namespace == 'molecular_function', p1_go))
for p in p1_go_mf:
assert ontology.id_to_node(p, dag).namespace == 'molecular_function'
p2_go_mf = set(filter(lambda x: ontology.id_to_node(x, dag).namespace == 'molecular_function', p2_go))
for p in p2_go_mf:
assert ontology.id_to_node(p, dag).namespace == 'molecular_function'
fp.write("{0}\t{1}\t{2}\t{3}\t{4}\t{5}\t{6}\t{7}\n".format(
p1, p2,
','.join(p1_go_cc), ','.join(p2_go_cc),
','.join(p1_go_bp), ','.join(p2_go_bp),
','.join(p1_go_mf), ','.join(p2_go_mf)
)
)
fp.close()
# Run R script then collect output from tmp file
args = [
'Rscript',
'semantic_sim.r',
'--file={}'.format(r_file_in),
'--out={}'.format(r_file_out)
]
proc = subprocess.Popen(args)
proc.wait()
# Parse the r_output into a list
sims_tuple = []
with open(r_file_out, 'r') as fp:
for line in fp:
xs = line.strip().split('\t')
p1, p2, cc_ss, bp_ss, mf_ss = xs
sims_tuple.append((p1, p2, cc_ss, bp_ss, mf_ss))
fp.close()
os.remove(r_file_in)
os.remove(r_file_out)
return sims_tuple
