def test_selection_writing(self):
"""Test record_has, and writerec.
Adapted from Bio.UniProt.GOA.py by Iddo Friedberg [email protected].
"""
recs = []
filtered = []
# Fields to filter
evidence = {"Evidence": {"ND"}}
synonym = {"Synonym": {"YA19A_YEAST", "YAL019W-A"}}
taxon_id = {"Taxon_ID": {"taxon:559292"}}
# Temporal file to test writerec
f_number, f_filtered = tempfile.mkstemp()
os.close(f_number)
# Open a file and select records as per filter
with open("UniProt/goa_yeast.gaf", "r") as handle:
for rec in GOA.gafiterator(handle):
recs.append(rec)
# Filtering
if (
GOA.record_has(rec, taxon_id)
and GOA.record_has(rec, evidence)
and GOA.record_has(rec, synonym)
):
filtered.append(rec)
# Check number of filtered records
self.assertEqual(len(filtered), 3)
# Write the filtered records to a file using writerec
with open(f_filtered, "w") as handle:
# '!gaf-version: 2.1'
handle.write("!gaf-version: 2.1 \n") # Adding file header
for rec in filtered:
GOA.writerec(rec, handle)
# Open and read the file containing the filtered records
recs_ff = [] # Records from filtered file
with open(f_filtered, "r") as handle:
for rec in GOA.gafiterator(handle):
recs_ff.append(rec)
# Delete test file
os.remove(f_filtered)
# Compare, recs saved by writerec and filtered recs
self.assertEqual(filtered, recs_ff)
def filter_in_experimental(handle):
outhandle = open(handle.name + ".exp_evidence", "w")
outhandle.write('!gaf-version: 2.0\n')
for inrec in upg.gafiterator(handle):
if upg.record_has(inrec, {'Evidence': GO_EXP_EC}):
upg.writerec(inrec, outhandle)
outhandle.close()
def has_experimental(goa_reclist):
retval = False
for rec in goa_reclist:
if upg.record_has(rec, {'Evidence': GO_EXP_EC}):
retval = True
break
return retval
def read_gaf_write_tab(gaf_file, include_mfo, outfile):
Evidence = {'Evidence': set(['EXP','IDA','IPI','IMP','IGI','IEP','TAS','IC'])}
if include_mfo:
Aspect = {'Aspect':set(['P','F'])}
else:
Aspect = {'Aspect':set(['P'])}
Evidence = {'Evidence': set(['EXP','IDA','IPI','IMP','IGI','IEP','TAS','IC'])}
outhandle = open(outfile, 'w')
ingafhandle = open(gaf_file,'r')
counter = 0
for rec in GOA.gafiterator(ingafhandle):
if GOA.record_has(rec, Aspect):
if GOA.record_has(rec, Evidence):
prot = rec['DB_Object_ID']
go = rec['GO_ID']
outhandle.write("%s\t%s\n" % (prot, go))
counter +=1
ingafhandle.close()
outhandle.close()
return(counter)
"""
Returns the pmids of the papers this paper cites
"""
cites_list = []
handle = ez.efetch("pubmed", id=pmid, retmode="xml")
pubmed_rec = ez.parse(handle).__next__()
for ref in pubmed_rec['MedlineCitation']['CommentsCorrectionsList']:
if ref.attributes['RefType'] == 'Cites':
cites_list.append(str(ref['PMID']))
return cites_list
f = open ("papers and citations.txt","w")
st = "GO-annotated proteins supported by IGI evidence (Inferred from Genetic Interaction)\n"
handle = open("gene_association.goa_yeast")
proteins = goa.gafiterator(handle)
Evi_Aspect = {"Evidence":set(["IGI"])}
for protein in proteins:
if goa.record_has(protein, Evi_Aspect):
for p in protein['DB:Reference']:
if p[:4] == "PMID":
st += "Main PubMed reference: "+ p +"\n"
citations = get_citations(p[5:])
for cit in citations:
st += cit + " "
st += "\n"
f.write(st)
f.close()
#!/usr/bin/env python
import sys
import argparse
import target_prep as tp
from Bio.UniProt import GOA as upg
if __name__ == '__main__':
# parser = argparse.ArgumentParser(description='Filter by field')
# parser.add_argument('-o','--output')
# parser.add_argument('-f','--field')
outhandle = sys.stdout
if len(sys.argv) == 5:
outhandle = open(sys.argv[4],"w")
outhandle.write('!gaf-version: 2.0\n')
goodvals = {sys.argv[1]: set(sys.argv[2].split(','))}
for inrec in upg.gafiterator(open(sys.argv[3])):
if upg.record_has(inrec, goodvals):
upg.writerec(inrec,outhandle)
"""""
Retrieving protein references from the yeast association file in GAF 2.0 format
according to different criteria
"""
""""
Retrieve all references cited to annotate proteins with Experimental Evidence Codes
"""
handle = open("gene_association.goa_yeast") # open the association gene file of the yeast
proteins = goa.gafiterator(handle) # read all records in the file
Evidences = {"Evidence":set(["EXP", "IDA", "IPI", "IMP", "IGI", "IEP"])}
print ("GO-annotated proteins supported by Experimental Evidence Code")
for protein in proteins:
if goa.record_has(protein, Evidences):
print(protein['DB:Reference'])
""""
Retrieve all references cited to annotate proteins with Experimental Evidence Codes
in the Molecular Function aspect of GO
"""
handle = open("gene_association.goa_yeast")
proteins = goa.gafiterator(handle)
Evi_Aspect = {"Evidence":set(["EXP", "IDA", "IPI", "IMP", "IGI", "IEP"]), "Aspect":set(["F"])}
print ("GO-annotated proteins supported by Experimental Evidence Code in the Molecular Function Ontology")
for protein in proteins:
if goa.record_has(protein, Evi_Aspect):
print(protein['DB:Reference'])
print("\n")