Browse files

Remove unused function in avian_flu_upload

  • Loading branch information...
trvrb committed Jan 3, 2019
1 parent 38a44e5 commit cb06bf3f031cbfc719f65bd69fa7ebe41b38f4e7
Showing with 0 additions and 30 deletions.
  1. +0 −30 vdb/
@@ -428,36 +428,6 @@ def determine_group_fields(self, v, patterns, **kwargs):
v['vtype'], v['subtype'], v['lineage'] = match[0], match[1], match[2]
return v

def align_flu(self, doc, min_score_percentage=0.85, **kwargs):
align with sequence from outgroup to determine subtype and lineage
:return: True if determined grouping, False otherwise
scores = []
from Bio.Seq import Seq
from Bio.SeqRecord import SeqRecord
from Bio.Alphabet import IUPAC
from Bio import AlignIO
record = SeqRecord(Seq(doc['sequence'],
for olineage, oseq in self.outgroups.items():
SeqIO.write([oseq, record], "temp_in.fasta", "fasta")
os.system("mafft --auto temp_in.fasta > temp_out.fasta 2>tmp")
tmp_aln = np.array('temp_out.fasta', 'fasta'))
scores.append((olineage, (tmp_aln[0]==tmp_aln[1]).sum()))
scores.sort(key = lambda x:x[1], reverse=True)
if scores[0][1]>min_score_percentage*len(record.seq):
print("Lineage based on similarity:", scores[0][0], doc['strain'], len(record.seq), scores)
return self.outgroup_patterns[scores[0][0]]
print("Couldn't parse virus subtype and lineage from aligning sequence: ", doc['strain'], len(record.seq), scores)
return None
print("Alignment failed: " + doc['strain'])
return None

if __name__=="__main__":
args = parser.parse_args()
sequence_fasta_fields = {0: 'accession', 1: 'strain', 2: 'isolate_id', 3:'locus', 4: 'passage', 5: 'submitting_lab'}

0 comments on commit cb06bf3

Please sign in to comment.