def _handle_blast_hit(self, in_file, database_name, blast_file, results,
hit_seq_records):
blast_table = pd.read_table(
blast_file,
header=None,
names=BlastHandler.BLAST_COLUMNS,
index_col=False).astype(dtype={
'qseqid': np.unicode_,
'sseqid': np.unicode_
})
partitions = BlastHitPartitions()
blast_table['plength'] = (blast_table.length /
blast_table.qlen) * 100.0
blast_table = blast_table[
(blast_table.pident >= self._pid_threshold)
& (blast_table.plength >= self._plength_threshold)
& ~blast_table.qseqid.isin(self._genes_to_exclude)]
blast_table.sort_values(by=self.BLAST_SORT_COLUMNS, inplace=True)
for index, blast_record in blast_table.iterrows():
partitions.append(
self._create_hit(in_file, database_name, blast_record))
for hits_non_overlapping in partitions.get_hits_nonoverlapping_regions(
):
for hit in self._select_hits_to_include(hits_non_overlapping):
blast_results = self._get_result_rows(hit, database_name)
if blast_results is not None:
logger.debug("record = %s", blast_results)
results.extend(blast_results)
hit_seq_records.append(hit.get_seq_record())
def testTwoPartitionsDifferentContigNames(self):
hit1 = AMRHitHSP(None, None)
hit1.get_genome_contig_id = MagicMock(return_value="contig1")
hit1.get_genome_contig_start = MagicMock(return_value=1)
hit1.get_genome_contig_end = MagicMock(return_value=10)
hit1.get_genome_contig_strand = MagicMock(return_value='plus')
hit2 = AMRHitHSP(None, None)
hit2.get_genome_contig_id = MagicMock(return_value="contig2")
hit2.get_genome_contig_start = MagicMock(return_value=1)
hit2.get_genome_contig_end = MagicMock(return_value=10)
hit2.get_genome_contig_strand = MagicMock(return_value='plus')
parts = BlastHitPartitions()
parts.append(hit1)
parts.append(hit2)
return_list = parts.get_hits_nonoverlapping_regions()
self.assertEqual(2, len(return_list), "Should be two partitions")
self.assertEqual(1, len(return_list[0]), "Partition 1 should have 1 hit")
self.assertEqual(['contig1'], [x.get_genome_contig_id() for x in return_list[0]],
"Should have correct contig names")
self.assertEqual([1], [x.get_genome_contig_start() for x in return_list[0]],
"Should have correct contig starts")
self.assertEqual([10], [x.get_genome_contig_end() for x in return_list[0]], "Should have correct contig ends")
self.assertEqual(1, len(return_list[1]), "Partition 2 should have 1 hit")
self.assertEqual(['contig2'], [x.get_genome_contig_id() for x in return_list[1]],
"Should have correct contig names")
self.assertEqual([1], [x.get_genome_contig_start() for x in return_list[1]],
"Should have correct contig starts")
self.assertEqual([10], [x.get_genome_contig_end() for x in return_list[1]], "Should have correct contig ends")
def testSinglePartitionHit2EdgeWithinHit1Greater(self):
hit1 = AMRHitHSP(None, None)
hit1.get_genome_contig_id = MagicMock(return_value="contig1")
hit1.get_genome_contig_start = MagicMock(return_value=5)
hit1.get_genome_contig_end = MagicMock(return_value=11)
hit1.get_genome_contig_strand = MagicMock(return_value='plus')
hit2 = AMRHitHSP(None, None)
hit2.get_genome_contig_id = MagicMock(return_value="contig1")
hit2.get_genome_contig_start = MagicMock(return_value=10)
hit2.get_genome_contig_end = MagicMock(return_value=15)
hit2.get_genome_contig_strand = MagicMock(return_value='plus')
parts = BlastHitPartitions()
parts.append(hit1)
parts.append(hit2)
return_list = parts.get_hits_nonoverlapping_regions()
self.assertEqual(1, len(return_list), "Should only be one partition")
self.assertEqual(2, len(return_list[0]), "Should be two hits")
self.assertEqual(['contig1', 'contig1'], [x.get_genome_contig_id() for x in return_list[0]],
"Should have correct contig names")
self.assertEqual([5, 10], [x.get_genome_contig_start() for x in return_list[0]],
"Should have correct contig starts")
self.assertEqual([11, 15], [x.get_genome_contig_end() for x in return_list[0]],
"Should have correct contig ends")
def testSinglePartitionMinus(self):
hit1 = AMRHitHSP(None, None)
hit1.get_genome_contig_id = MagicMock(return_value="contig1")
hit1.get_genome_contig_start = MagicMock(return_value=10)
hit1.get_genome_contig_end = MagicMock(return_value=1)
hit1.get_genome_contig_strand = MagicMock(return_value='minus')
parts = BlastHitPartitions()
parts.append(hit1)
return_list = parts.get_hits_nonoverlapping_regions()
self.assertEqual(1, len(return_list), "Should only be one partition")
self.assertEqual(1, len(return_list[0]), "Should only be one hit")
self.assertEqual('contig1', return_list[0][0].get_genome_contig_id(), "Should have correct contig name")
self.assertEqual(10, return_list[0][0].get_genome_contig_start(), "Should have correct contig start")
self.assertEqual(1, return_list[0][0].get_genome_contig_end(), "Should have correct contig end")
def testSinglePartitionPlusFailMinusCoords(self):
hit1 = AMRHitHSP(None, None)
hit1.get_genome_contig_id = MagicMock(return_value="contig1")
hit1.get_genome_contig_start = MagicMock(return_value=10)
hit1.get_genome_contig_end = MagicMock(return_value=1)
hit1.get_genome_contig_strand = MagicMock(return_value='plus')
parts = BlastHitPartitions()
self.assertRaises(InvalidPositionException, parts.append, hit1)
def testSinglePartitionIdenticalHitsMinusStrand(self):
hit1 = AMRHitHSP(None, None)
hit1.get_genome_contig_id = MagicMock(return_value="contig1")
hit1.get_genome_contig_start = MagicMock(return_value=10)
hit1.get_genome_contig_end = MagicMock(return_value=1)
hit1.get_genome_contig_strand = MagicMock(return_value='minus')
parts = BlastHitPartitions()
parts.append(hit1)
parts.append(hit1)
return_list = parts.get_hits_nonoverlapping_regions()
self.assertEqual(1, len(return_list), "Should only be one partition")
self.assertEqual(2, len(return_list[0]), "Should be two hits")
self.assertEqual(['contig1', 'contig1'], [x.get_genome_contig_id() for x in return_list[0]],
"Should have correct contig names")
self.assertEqual([10, 10], [x.get_genome_contig_start() for x in return_list[0]],
"Should have correct contig starts")
self.assertEqual([1, 1], [x.get_genome_contig_end() for x in return_list[0]],
"Should have correct contig ends")