def go_terms_with_ec_per_paper(papers,outpath=None,top=20):
"""Create a dict that counts up how many times a specific (GO ID, GO Term Text, EvCode)
tuple occurs for each paper"""
#
# Can be used with SP & GOA data
go_ec_count = {}
go_con = mysqlConnect()
go_cur = go_con.cursor()
for p in papers:
for rec in papers[p]:
go_id = rec['go_id']
go_ec = rec['go_ec']
try:
name = gu.go_acc_to_name(go_id,go_cur)
except IndexError: #sometimes the GO ID given is actually a synonym
try:
name = gu.go_acc_to_synonym_name(go_id, go_cur)
except IndexError: #sometimes it just doesn't work
print "problem with GO ID", go_id
name = ''
gokey = (go_id, name, go_ec)
# go_ec_count[PMID] = {{(GO ID, GO Term Text, Ev Code) : # times paper gives this annotaion}}
if p in go_ec_count:
go_ec_count[p][gokey] = go_ec_count[p].get(gokey,0) + 1
else:
go_ec_count[p] = {gokey: 1}
go_con.close()
return go_ec_count
def go_terms_with_ec_per_paper(papers, outpath=None, top=20):
"""Create a dict that counts up how many times a specific (GO ID, GO Term Text, EvCode)
tuple occurs for each paper"""
#
# Can be used with SP & GOA data
go_ec_count = {}
go_con = mysqlConnect()
go_cur = go_con.cursor()
for p in papers:
for rec in papers[p]:
go_id = rec['go_id']
go_ec = rec['go_ec']
try:
name = gu.go_acc_to_name(go_id, go_cur)
except IndexError: #sometimes the GO ID given is actually a synonym
try:
name = gu.go_acc_to_synonym_name(go_id, go_cur)
except IndexError: #sometimes it just doesn't work
print "problem with GO ID", go_id
name = ''
gokey = (go_id, name, go_ec)
# go_ec_count[PMID] = {{(GO ID, GO Term Text, Ev Code) : # times paper gives this annotaion}}
if p in go_ec_count:
go_ec_count[p][gokey] = go_ec_count[p].get(gokey, 0) + 1
else:
go_ec_count[p] = {gokey: 1}
go_con.close()
return go_ec_count
def go_terms_per_paper(papers,outpath=None,top=20):
#
# Can be used with SP & GOA data
go_count = {}
go_con = mysqlConnect()
go_cur = go_con.cursor()
for p in papers:
for rec in papers[p]:
go_id = rec['go_id']
try:
name = gu.go_acc_to_name(go_id,go_cur)
except IndexError: #sometimes the GO ID given is actually a synonym
try:
name = gu.go_acc_to_synonym_name(go_id, go_cur)
except IndexError: #sometimes it just doesn't work
print "problem with GO ID", go_id
name = ''
gokey = (go_id, name)
# go_count[PMID] = {{(GO ID, GO Term Text) : # times paper gives this annotaion}}
if p in go_count:
go_count[p][gokey] = go_count[p].get(gokey,0) + 1
else:
go_count[p] = {gokey: 1}
go_con.close()
return go_count
def go_terms_per_paper(papers, outpath=None, top=20):
#
# Can be used with SP & GOA data
go_count = {}
go_con = mysqlConnect()
go_cur = go_con.cursor()
for p in papers:
for rec in papers[p]:
go_id = rec['go_id']
try:
name = gu.go_acc_to_name(go_id, go_cur)
except IndexError: #sometimes the GO ID given is actually a synonym
try:
name = gu.go_acc_to_synonym_name(go_id, go_cur)
except IndexError: #sometimes it just doesn't work
print "problem with GO ID", go_id
name = ''
gokey = (go_id, name)
# go_count[PMID] = {{(GO ID, GO Term Text) : # times paper gives this annotaion}}
if p in go_count:
go_count[p][gokey] = go_count[p].get(gokey, 0) + 1
else:
go_count[p] = {gokey: 1}
go_con.close()
return go_count
