def test_overlapping_deletion(self):
graph = graphContainer.GraphContainer()
n = graph.add_altNode("chr", 10, 17, "ATCGATCG")
varDel = {"start": 2, "end": 4, "alt": ""}
varSNV = {"start": 4, "end": 4, "alt": "C"}
variants.add_variants(graph, {n['name']: [varDel, varSNV]})
graphUtils.remove_empty_nodes(graph)
nodeNames = [n['name'] for n in graph.nodes.values()]
self.assertCountEqual(nodeNames, [
"chr:10-11:AT", "chr:12-13:CG", "chr:14-14:A", "chr:14-14:C",
"chr:15-17:TCG"
])
left = graph.nodes["chr:10-11:AT"]
right = graph.nodes["chr:15-17:TCG"]
alt = graph.nodes["chr:14-14:C"]
ref = graph.nodes["chr:14-14:A"]
ins = graph.nodes["chr:12-13:CG"]
self.assertTrue(graph.has_edge(left, right))
self.assertTrue(graph.has_edge(left, ins))
self.assertTrue(graph.has_edge(ins, ref))
self.assertTrue(graph.has_edge(ref, right))
self.assertTrue(graph.has_edge(ins, alt))
self.assertTrue(graph.has_edge(alt, right))
self.assertFalse(graph.has_edge(left, ref))
self.assertFalse(graph.has_edge(left, alt))
self.assertFalse(graph.has_edge(ref, ins))
def test_ins_end(self):
graph = graphContainer.GraphContainer()
r = graph.add_refNode("chr", 18, 20)
n = graph.add_altNode("chr", 10, 17, "ATCGATCG")
graph.add_edge(n, r, ["foo"])
var = {"start": 8, "end": 7, "alt": "CCC"}
variants.add_variants(graph, {n['name']: [var]})
graphUtils.remove_empty_nodes(graph)
left = graph.nodes["chr:10-17:ATCGATCG"]
right = graph.nodes["ref-chr:18-20"]
alt = graph.nodes["chr:18-17:CCC"]
self.assertEqual(len(graph.nodes), 3)
self.assertTrue(graph.has_edge(left, right))
self.assertTrue(graph.has_edge(left, alt))
self.assertTrue(graph.has_edge(alt, right))
self.assertFalse(graph.has_edge(alt, left))
self.assertFalse(graph.has_edge(right, alt))
self.assertCountEqual(graph.get_edge(left['name'], right['name'])['sequences'], ["foo"])
self.assertCountEqual(graph.get_edge(alt['name'], right['name'])['sequences'], ["foo"])
def run(args):
levels = [logging.WARNING, logging.INFO, logging.DEBUG]
logging.basicConfig(format='%(message)s',
level=levels[min(args.verbose,
len(levels) - 1)])
graphDict = load_json(args.graph)
graph = graphUtils.load_json(graphDict)
if args.variants:
varJson = load_json(args.variants)
if "variants" not in varJson:
raise Exception("No variants in variant JSON")
varDict = varJson["variants"]
else:
varDict = graphDict.get("variants", {})
if not varDict:
logging.warning("No variants in graph")
print(varDict.keys())
variants.add_variants(graph, varDict)
graphUtils.remove_empty_nodes(graph)
json.dump(graph.json_dict(), args.output, sort_keys=True)
def test_var_begin(self):
graph = graphContainer.GraphContainer()
r = graph.add_refNode("chr", 1, 9)
n = graph.add_altNode("chr", 10, 17, "ATCGATCG")
graph.add_edge(r, n, ["foo"])
var = {"start": 0, "end": 0, "alt": "G"}
variants.add_variants(graph, {n['name']: [var]})
graphUtils.remove_empty_nodes(graph)
left = graph.nodes["ref-chr:1-9"]
right = graph.nodes["chr:11-17:TCGATCG"]
ref = graph.nodes["chr:10-10:A"]
alt = graph.nodes["chr:10-10:G"]
self.assertEqual(len(graph.nodes), 4)
self.assertTrue(graph.has_edge(left, ref))
self.assertTrue(graph.has_edge(left, alt))
self.assertTrue(graph.has_edge(ref, right))
self.assertTrue(graph.has_edge(alt, right))
self.assertFalse(graph.has_edge(left, right))
self.assertFalse(graph.has_edge(ref, alt))
self.assertCountEqual(graph.get_edge(left['name'], ref['name'])['sequences'], ["foo"])
self.assertCountEqual(graph.get_edge(left['name'], alt['name'])['sequences'], ["foo"])
def convert_vcf(vcf,
ref,
target_regions=None,
ref_node_padding=150,
ref_node_max_length=1000,
allele_graph=False,
simplify=True,
alt_paths=False,
alt_splitting=False):
"""
Convert a single VCF file to a graph dictionary
:param vcf: file name of the VCF file
:param ref: reference FASTA file name
:param target_regions: target region list
:param ref_node_padding: padding / read length
:param ref_node_max_length: maximum length before splitting a reference node
:param allele_graph: add edges between any compatible allele pair, not just haplotypes from input
:param simplify: simplify the graph
:param alt_paths: Add all possible non-reference paths to the graph
:param alt_splitting: also split long alt nodes (e.g. long insertions)
:return: dictionary containing JSON graph
"""
graph = GraphContainer("Graph from %s" % vcf)
indexed_vcf = tempfile.NamedTemporaryFile(delete=False, suffix=".vcf.gz")
try:
indexed_vcf.close()
# noinspection PyUnresolvedReferences
pysam.bcftools.view(vcf, "-o", indexed_vcf.name, "-O", "z", catch_stdout=False) # pylint: disable=no-member
# noinspection PyUnresolvedReferences
pysam.bcftools.index(indexed_vcf.name) # pylint: disable=no-member
regions = map(parse_region, target_regions) if target_regions else [(None,)*3]
for (chrom, start, end) in regions:
if chrom is not None:
logging.info(f"Starting work on region: {chrom}:{start}-{end}")
try:
vcfGraph = VCFGraph.create_from_vcf(
ref, indexed_vcf.name, chrom, start, end, ref_node_padding, allele_graph)
except NoVCFRecordsException:
logging.info(f"Region {chrom}:{start}-{end} has no VCF records, skipping.")
continue
logging.info(f"CONSTRUCTED VCF GRAPH:\n{str(vcfGraph)}")
chromGraph = vcfGraph.get_graph(allele_graph)
if ref_node_max_length:
graphUtils.split_ref_nodes(chromGraph, ref_node_max_length, ref_node_padding)
if alt_splitting:
graphUtils.split_alt_nodes(chromGraph, ref_node_max_length, ref_node_padding)
if simplify:
graphUtils.remove_empty_nodes(chromGraph)
graphUtils.combine_nodes(chromGraph)
# Disable edge label simplification for now. May use node-label short-cut later
# graphUtils.remove_redundant_edge_labels(graph)
chromGraph.check()
graphUtils.add_graph(graph, chromGraph)
finally:
os.remove(indexed_vcf.name)
graph.target_regions = target_regions or graph.get_reference_regions()
graphUtils.add_source_sink(graph)
graphUtils.add_ref_path(graph)
if alt_paths:
graphUtils.add_alt_paths(graph)
graph.check()
return graph.json_dict()