def _search(self, taxids):
"""Search GenBank for matches, increasing thoroughness if
fewer matches than target nseqs"""
# record seqids of all sequences that match query
# move to next thoroughness if too few sequences
# if _search is run again, will not return seqids of sequences
# that have already been found -- self.deja_vues records seen
# sequences
# limit number of taxids to 5, search recursively if more than 5
if len(taxids) > 5:
res = self._search(taxids[:5])
self.thoroughness = 1
res.extend(self._search(taxids[5:]))
return res
seqids = []
while len(seqids) < self.nseqs:
if self.thoroughness > self.max_thoroughness:
# keep searching until enough sequences have been
# found or until the max thoroughness has been hit
break
search_term = self._buildSearchTerm(taxids, self.thoroughness)
seqcount = etools.eSearch(search_term, logger=self.logger)['Count']
if int(seqcount) >= 1:
# return ALL matching seqids if more than 0
seqids.extend(etools.eSearch(search_term, logger=self.logger,
retMax=seqcount)['IdList'])
# filter those that have already been seen
seqids = [e for e in seqids if e not in self.deja_vues]
self.thoroughness += 1
self.deja_vues.extend(seqids)
self.deja_vues = list(set(self.deja_vues))
return list(set(seqids))
def getOutgroup(namesdict, parentid, logger, outgroupid=None, minrecords=1000):
"""Return namesdict with suitable outgroup"""
# TODO: too complex, consider breaking up
def findParent(parentid):
return etools.eFetch(parentid, logger=logger,
db="taxonomy")[0]['ParentTaxId']
def getTaxIdMetaData(ncbi_id):
etal_bool = False
if len(ncbi_id) > 1:
ncbi_id = ncbi_id[0]
etal_bool = True
record = etools.eFetch(ncbi_id, logger=logger, db="taxonomy")[0]
metadata = [record['Rank'], record['ScientificName']]
if etal_bool:
metadata = [e + ' et al.' for e in metadata]
return metadata[0], metadata[1]
# loop until a suitable outgroup is found. Criteria are:
# 1. ids returned must belong to a sister group of all ids of
# names given
# 2. ids must have nucleotide data (i.e.avoid returning extinct organisms)
# assumptions:
# 1. NCBI taxonomy is not paraphyletic
# make sure parentid is string
if not outgroupid:
parentid = str(parentid)
outgroup_ids = []
while not outgroup_ids:
# if parent id are Cellular Orgs, likely name resolution error
# or names given are too diverse
if parentid == '131567':
raise TaxonomicRankError()
# get parent of parent
grandparentid = findParent(parentid)
# find all children
candidates = etools.findChildren(grandparentid, logger=logger,
next=True)
# filter out children that are in ingroup
candidates = [e for e in candidates if e != parentid]
# search genbank for nuc records
for candidate in candidates:
term = 'txid' + str(candidate) + '[PORGN]'
nuc_record = etools.eSearch(term, logger=logger)
# there must be more than 1000 nuc records
if int(nuc_record['Count']) > minrecords:
outgroup_ids.append(candidate)
# make grandparentid the new parentid
parentid = grandparentid
else:
outgroup_ids = [outgroupid]
# add outgroup_ids to namesdict
rank, unique_name = getTaxIdMetaData(outgroup_ids)
# convert to ints
outgroup_ids = [int(e) for e in outgroup_ids]
namesdict["outgroup"] = {"txids": outgroup_ids, "unique_name": unique_name,
"rank": rank}
return namesdict
