def all_exclusive_IEA(handle):
outhandle = open(handle.name + ".exclusive_IEA", "w")
outhandle.write('!gaf-version: 2.0\n')
for protrec in upg.gafbyproteiniterator(handle):
if exclusive_IEA(protrec):
for outrec in protrec:
upg.writerec(outrec, outhandle)
outhandle.close()
def all_hasnt_experimental(handle):
outhandle = open(handle.name + ".noexp", "w")
outhandle.write('!gaf-version: 2.0\n')
for protrec in upg.gafbyproteiniterator(handle):
if not has_experimental(protrec):
for outrec in protrec:
upg.writerec(outrec, outhandle)
outhandle.close()
def became_experimental(handle_iea, handle_exp):
"""Identify Electronically annotated proteins that became experimentally
annotated later
handle_iea: gaf file with proteins exclusively IEA annotated
handle_exp: gaf file with proteins experimentally annotated
Note: files should contain only one ontology (either MFO, BPO or
CCO) to make sense.
"""
expdict = {}
outhandle = open(handle_exp.name + ".became_exp", "w")
outhandle.write('!gaf-version: 2.0\n')
# First read experimental into memory
for exprec in upg.gafbyproteiniterator(handle_exp):
expdict[exprec[0]['DB_Object_ID']] = exprec
# Now read in the non-experimental
for iearec in upg.gafbyproteiniterator(handle_iea):
prot_id = iearec[0]['DB_Object_ID']
if prot_id in expdict:
upg.writebyproteinrec(expdict[prot_id], outhandle)
outhandle.close()
def species_stats(handle):
"""Statistics for species distributions in a gaf file"""
taxa_count = defaultdict(int)
for prot_rec in upg.gafbyproteiniterator(handle):
taxa_count[prot_rec[0]['Taxon_ID'][0]] += 1
return taxa_count
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")
""""
Retrieve all references cited to support the annotation of three UniProt
proteins whose IDs are: O13516, YPR010C-A, STE2
"""
handle = open("gene_association.goa_yeast")
proteins = list(goa.gafbyproteiniterator(handle)) # get all records in yeast file groupped by object id
IDs = {"DB_Object_ID":set(["A2P2R3","P00045","P50101"])}
for protein in proteins:
if goa.record_has(protein[0], IDs):
print("Protein ID:" + protein[0]["DB_Object_ID"]+"\n"+"References:")
for p in protein:
print(p['DB:Reference'])
print("\n")