def search(self):
if self.database == 'PubMed':
from Bio import PubMed
from Bio import GenBank
searchIds = PubMed.search_for(self.searchTerm, max_ids=self.maxResults)
GBrecParser = GenBank.FeatureParser()
ncbiDict = GenBank.NCBIDictionary(self.type,
'genbank',
parser=GBrecParser)
from Bio import Medline
MLrecParser = Medline.RecordParser()
medlineDict = PubMed.Dictionary(delay=1.0, parser=MLrecParser)
for id in searchIds:
MLrecord = medlineDict[id]
GBrecord = ncbiDict[id]
newDBItem = DBItem(self.project,
seq=GBrecord.seq,
descript=GBrecord.description,
id=id,
record=MLrecord)
self.items[id] = newDBItem
org = rec.annotations.get('organism', '')
date = rec.annotations.get('date', '')
head = '>gi:%s, id:%s, org:%s, date:%s\n' % (gi, rec.id, org, date)
body = '\n'.join(textwrap.wrap(rec.seq.data, width=80))
return head, body
if __name__ == '__main__':
mode = sys.argv[1]
text = sys.argv[2]
output_file = sys.argv[3]
print('Searching for %s <br>' % text)
# check if inputs are all numbers
try:
gi_list = text.split()
[int(_) for _ in gi_list]
except ValueError:
gi_list = GenBank.search_for(text, max_ids=10)
fp = open(output_file, 'wt')
record_parser = GenBank.FeatureParser()
ncbi_dict = GenBank.NCBIDictionary(mode, 'genbank', parser=record_parser)
for gid in gi_list:
res = ncbi_dict[gid]
head, body = make_fasta(res)
fp.write(head + body + '\n')
print(head)
fp.close()
def run(self):
if not self.allowRefSeqs:
print 'NOT ALLOWING REFSEQS'
if self.query_string.startswith(
'GI:') or self.query_string.startswith('gi:'):
self.query_string = self.query_string[3:]
q = self.query_string
gi_list = self.search(q)
else:
q = "mycobacterium phage " + self.query_string + " AND Hatfull GF[AUTH] NOT srcdb_refseq[prop]"
print "search query:", q
gi_list = self.search(q)
print 'gi_list:', gi_list
if len(gi_list) == 0:
print 'Got no results. Changing search criteria'
q = self.query_string + " AND Hatfull GF[AUTH] NOT srcdb_refseq[prop]"
print "search query:", q
gi_list = self.search(q)
if len(gi_list) == 0:
print 'Got no results. Changing search criteria'
q = self.query_string + " NOT srcdb_refseq[prop]"
print "search query:", q
gi_list = self.search(q)
if len(gi_list) != 0:
print 'found GenBank Direct Submission(s)'
print gi_list
else:
print 'found no results other than refSeq(s), which you refused'
self.result = None
return
else: # allowing refSeqs
print 'ALLOWING REFSEQS'
if self.query_string.startswith(
'GI:') or self.query_string.startswith('gi:'):
self.query_string = self.query_string[3:]
q = self.query_string
gi_list = self.search(q)
else:
q = "mycobacterium phage " + self.query_string + " AND Hatfull GF[AUTH]"
print "search query:", q
gi_list = self.search(q)
if len(gi_list) == 0:
q = self.query_string + " AND Hatfull GF[AUTH]"
gi_list = self.search(q)
if len(gi_list) == 0:
print 'Got no results. Changing search criteria'
print 'search query:', self.query_string
gi_list = self.search(self.query_string)
if len(gi_list) == 0:
print 'no results found'
self.results = gi_list
return
if len(gi_list) > 1:
selection = -1
for i in range(len(gi_list)):
print i + 1, '\t', gi_list[i]
selection = raw_input(
"Your search returned multiple results. Please type the number for your selection: "
)
selection = int(selection) - 1
else:
selection = 0
print 'creating parser...'
feature_parser = GenBank.FeatureParser()
print 'creating dict'
ncbi_dict = GenBank.NCBIDictionary('nucleotide',
'genbank',
parser=feature_parser)
if selection == -1: ## Accounts for non-existent phage query
print 'non-existent phage query'
self.result = 0
else:
print 'got result'
self.result = ncbi_dict[gi_list[selection]]
