def create_linear_peaks_from_bed(linear_sequence_fasta_file, peaks_bed_file,
obg_graph_file_name, vg_graph_file_name,
start_node, region):
ob_graph = obg.GraphWithReversals.from_file(obg_graph_file_name)
search_sequence = open(linear_sequence_fasta_file).read()
sequence_retriever = SequenceRetriever.from_vg_graph(vg_graph_file_name)
traverser = GraphTraverserUsingSequence(ob_graph, search_sequence,
sequence_retriever)
traverser.search_from_node(start_node)
linear_path_interval = traverser.get_interval_found()
IntervalCollection([linear_path_interval
]).to_file("linear_path.intervalcollection",
text_file=True)
print("Length")
print(linear_path_interval.length())
print(linear_path_interval.region_paths[0])
print(linear_path_interval.start_position)
print(linear_path_interval.end_position)
linear_peaks = PeakCollection.create_from_linear_intervals_in_bed_file(
obg_graph_file_name, linear_path_interval, peaks_bed_file,
region.start, region.end)
linear_peaks.to_file("linear_peaks.intervalcollection", text_file=True)
class TestSequences(unittest.TestCase):
def _test_from_vg_graph(self):
retriever = SequenceRetriever.from_vg_graph("cactus-mhc.vg")
def setUp(self):
self.nodes = {1: "AAG", 2: "GAA", 3: "AGA"}
self.retriever = SequenceRetriever(self.nodes)
def test_single_node_interval(self):
interval = Interval(0, 3, [1])
self.assertEqual(self.retriever.get_interval_sequence(interval),
self.nodes[1])
def test_single_node_interval_with_offset(self):
interval = Interval(1, 3, [1])
self.assertEqual(self.retriever.get_interval_sequence(interval), "AG")
def test_single_node_interval_with_dual_offset(self):
interval = Interval(1, 2, [1])
self.assertEqual(self.retriever.get_interval_sequence(interval), "A")
def test_reversed_single_node_interval(self):
interval = Interval(0, 3, [-1])
self.assertEqual(self.retriever.get_interval_sequence(interval), "CTT")
def test_reversed_single_node_interval_with_dual_offsetl(self):
interval = Interval(1, 2, [-3])
self.assertEqual(self.retriever.get_interval_sequence(interval), "C")
def test_multiple_nodes_interval(self):
interval = Interval(0, 3, [1, 2])
self.assertEqual(self.retriever.get_interval_sequence(interval),
"AAGGAA")
def test_multiple_nodes_interval_second_rp_reversed(self):
interval = Interval(0, 3, [1, -2])
self.assertEqual(self.retriever.get_interval_sequence(interval),
"AAGTTC")
def test_long_interval(self):
interval = Interval(1, 1, [1, 2, 3, -3, -2, -1])
self.assertEqual(self.retriever.get_interval_sequence(interval),
"AGGAAAGATCTTTCC")
def test_from_vg_json_graph(self):
retriever = SequenceRetriever.from_vg_json_graph(
"tests/simple_graph.json")
self.assertEqual(retriever.get_sequence(1, 0, 7), "tttcccc")
def write_sequence_and_intervals(self, n=100):
# obg.IntervalCollection(self.intervals).to_file(
# "simulated_intervals.py")
logging.info("Getting sequences")
self.retriever = SequenceRetriever.from_vg_graph(
"../tests/haplo1kg50-mhc.vg")
sequences = [self.retriever.get_interval_sequence(i)
for i in self.intervals]
with open("simulated_sequences.fq", "w") as f:
for i, seq in enumerate(sequences):
f.write("@sim" + str(i) + "\n")
f.write(seq + "\n")
f.write("+\n")
f.write("~"*36 + "\n")
def find_linear_path_through_chromosome(chromosome, chromend, fasta_file_name,
ob_graph_file_name,
vg_graph_file_name):
genome = Fasta(fasta_file_name)
seq = str(genome[chromosome][0:50818468]).lower()
logging.info("Creating sequence retriever")
sequence_retriever = SequenceRetriever.from_vg_json_graph(
vg_graph_file_name)
graph = GraphWithReversals.from_numpy_file(ob_graph_file_name)
start_nodes = graph.get_first_blocks()
assert len(start_nodes) == 1, "Found %d start nodes" % start_nodes
start_node = start_nodes[0]
traverser = GraphTraverserUsingSequence(graph, seq, sequence_retriever)
traverser.search_from_node(start_node)
path = traverser.get_interval_found()
path = IntervalCollection(path)
path.to_file("22_path.intervalcollection", text_file=True)
logging.info("Done")
def _create_data(self):
node_offset = 1
for chrom_number, chromosome in enumerate(self.chromosomes):
graph = Graph(
{i + node_offset: Block(10)
for i in range(0, 3)},
{i + node_offset: [i + 1 + node_offset]
for i in range(0, 2)})
linear_map = LinearMap.from_graph(graph)
linear_map_file_name = "linear_map_%s.npz" % chromosome
linear_map.to_file(linear_map_file_name)
self.linear_maps.append(linear_map_file_name)
self.sequence_retrievers.append(
SequenceRetriever(
{i + node_offset: "A" * 10
for i in range(0, 3)}))
self._create_reads(chrom_number, chromosome, graph)
node_offset += 3
graph.convert_to_numpy_backend()
SequenceGraph.create_empty_from_ob_graph(graph).to_file(
chromosome + ".nobg.sequences")
graph.to_file(chromosome + ".nobg")
def _test_from_vg_graph(self):
retriever = SequenceRetriever.from_vg_graph("cactus-mhc.vg")
def test_from_vg_json_graph(self):
retriever = SequenceRetriever.from_vg_json_graph(
"tests/simple_graph.json")
self.assertEqual(retriever.get_sequence(1, 0, 7), "tttcccc")
def setUp(self):
self.nodes = {1: "AAG", 2: "GAA", 3: "AGA"}
self.retriever = SequenceRetriever(self.nodes)