def _stranded_ends(self, hit: AMRHitHSP) -> Tuple[int, int]:
"""
Gets the start/end coordinates, taking into account the strand.
:param hit: The hit.
:return: The (start,end) as a tuple.
"""
start = hit.get_genome_contig_start() if hit.get_genome_contig_strand() == 'plus' else hit.get_genome_contig_end()
end = hit.get_genome_contig_end() if hit.get_genome_contig_strand() == 'plus' else hit.get_genome_contig_start()
return start, end
def append(self, hit: AMRHitHSP) -> None:
"""
Adds a new blast hit to the set of partitions.
:param hit: The hit to add.
:return: None
"""
if hit.get_genome_contig_start() > hit.get_genome_contig_end() and hit.get_genome_contig_strand() == 'plus':
raise InvalidPositionException(
"Unsupported condition: strand=plus and contig start > contig end for hit (contig=" + hit.get_genome_contig_id() + ", start=" +
str(hit.get_genome_contig_start()) + ", end=" + str(hit.get_genome_contig_end()) + ")")
partition = self._get_existing_partition(hit)
if (partition is None):
self._create_new_parition(hit)
else:
self._add_hit_partition(hit, partition)
def _get_existing_partition(self, hit: AMRHitHSP) -> Optional[Dict[str, Any]]:
partition_name = hit.get_genome_contig_id()
if partition_name in self._partitions:
contig_partitions_list = self._partitions[partition_name]
for partition in contig_partitions_list:
if self._hit_in_parition(hit, partition):
return partition
return None
def _create_new_parition(self, hit: AMRHitHSP) -> None:
start, end = self._stranded_ends(hit)
contig_name = hit.get_genome_contig_id()
partition = {'start': start, 'end': end, 'hits': [hit]}
if contig_name in self._partitions:
self._partitions[contig_name].append(partition)
else:
self._partitions[contig_name] = [partition]
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 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 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")
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")