def test_blast_record_set(self):
# prepare database
seqfile_ops.write_fasta(self.db_file, self.db_records)
db_records_list = seqfile_ops.load_multifasta(self.db_file)
index = 0
for record in db_records_list:
self.assertEqual(record.id,self.db_records[index].id)
self.assertEqual(str(record.seq),str(self.db_records[index].seq))
index +=1
# make database
self.dbfile_path, db_report = blasting.make_blastDB(self.temp_dir,
self.db_name,
self.db_file,
'nucl')
self.assertIs(db_report['status'], 0)
self.assertEquals(db_report['message'], 'database exists')
# run local blast batch (with multiple queries)
matches_multi = blasting.blast_record_set(self.dbfile_path,
self.multi_records,
self.prefs)
self.assertIs(len(matches_multi), 3)
index = 0
for record in self.multi_records:
self.assertEqual(matches_multi[record.id][0]['contig_id'],
self.multi_records[index].id)
self.assertEqual(matches_multi[record
.id][0]['details']['match_p100'], 100)
index +=1
def genome_sets_load(genomes_path, input_file, input_prefs, db_path):
"""Load genome datasets listed in an input file."""
import os, sys
from classes.analysis_obj import GenomeSet
from analysis.seqfile_ops import ensure_fasta
from analysis.text_manipulation import adaptive_list_load
from analysis.blasting import make_blastDB
header = input_prefs['header']
columns = input_prefs['columns']
genomes_list = adaptive_list_load(input_file, header, columns)
print "prepping BLAST databases"
genome_sets = []
for line in genomes_list:
genome_name = line[0]
seq_file = os.path.join(genomes_path, line[1])
try: db_infile = ensure_fasta(seq_file)
except: raise
else: print "genome FASTA sequence available in", db_infile
dbfile_path, DB_report = make_blastDB(db_path, genome_name,
seq_file, 'nucl')
if DB_report['status'] is 1:
print genome_name, ":", DB_report['message']['error']
sys.exit()
elif DB_report['status'] is 0:
print genome_name, ":", DB_report['message']
new_genome_set = GenomeSet(db_infile, genome_name)
genome_sets.append(new_genome_set)
print " ", len(genome_sets),"databases ready to search"
return genome_sets
def test_local_blastn(self):
# prepare query
seqfile_ops.write_fasta(self.single_q_file, self.single_record)
query_record = seqfile_ops.load_fasta(self.single_q_file)
self.assertEqual(query_record.id,self.record_1.id)
self.assertEqual(str(query_record.seq),str(self.record_1.seq))
# prepare database
seqfile_ops.write_fasta(self.db_file, self.db_records)
records_list = seqfile_ops.load_multifasta(self.db_file)
index = 0
for record in records_list:
self.assertEqual(record.id,self.db_records[index].id)
self.assertEqual(str(record.seq),str(self.db_records[index].seq))
index +=1
# make database
self.dbfile_path, db_report = blasting.make_blastDB(self.temp_dir,
self.db_name,
self.db_file,
'nucl')
self.assertIs(db_report['status'], 0)
self.assertEquals(db_report['message'], 'database exists')
# run local blast with single query
self.status = blasting.local_blastn(self.single_q_file,
self.single_out_file,
self.dbfile_path,
self.prefs)
self.assertEquals(self.status['output'], '')
self.assertIsNone(self.status['error'])
# parse blast output
matches_single = blasting.parse_blast_out6(self.single_out_file,
self.prefs)
self.assertIs(len(matches_single), 1)
self.assertEqual(matches_single[0]['contig_id'],
self.single_record.id)
self.assertEqual(matches_single[0]['details']['match_p100'], 100)
