def run(self):
edges_path = cluster_svs.ClusterSVsStep(self.options).outpaths(final=True)["edges"]
clusters = pandas.read_table(edges_path)
clusters["chromx"] = clusters["chromx"].astype("string")
clusters["chromy"] = clusters["chromy"].astype("string")
assembled = []
utilities.ensure_dir(self.outpaths(final=False)["graphs"])
for cluster_number, cluster in clusters.groupby("cluster"):
self.logger.log(cluster_number)
try:
cur_assembled = self.walk(cluster_number, cluster)
assembled.append(cur_assembled)
except IOError:
print "not found", cluster_number
# TODO: deal with empty list
# TODO: normalize coordinates according to reference.compare_chroms()
assembled = pandas.concat(assembled, ignore_index=True)
assembled["x"] = assembled["x"].astype(int)
assembled["y"] = assembled["y"].astype(int)
print self.options.reference.chroms
print assembled["chromx"].unique()
print assembled["chromy"].unique()
outpath = self.outpaths(final=False)["walk_assemblies"]
assembled.to_csv(outpath, sep="\t", index=False)
def run(self):
self.assembly_dir = self.outpaths(final=False)["assembly_dir"]
utilities.ensure_dir(self.assembly_dir)
# TODO: how many reads is reasonable here?
max_reads = 5e6
fasta_path = self.combine_fastas(max_reads)
contigs_path = self.run_assembly(fasta_path)
self.align_contigs(contigs_path)
def visualize_graphs(outdir, graphs, evidence, file_label=""):
try:
utilities.ensure_dir(outdir)
supported = 0
missing = 0
breakpoints = 0
for i, graph in enumerate(graphs):
print "visualize", i, graph
graph = graph.copy()
for n1,n2,data in graph.edges(data=True):
data["label"] = "{}/{}={:.2g};{:.2g}".format(
int(data["shared"]),
int(data["total"]),
data["shared"]/float(data["total"]),
data["p"])
if data["kind"]=="facing":
data["label"] = "[{}]".format(data["label"])
data["style"] = "dashed"
elif data["kind"] == "breakpoint":
breakpoints += 1
elif data["kind"] == "weak":
data["fontsize"] = 11
data["color"] = "gray"
if "assembled" in data:
data["color"] = "orange"
for node in graph.nodes():
graph.node[node]["label"] = get_node_label(node)
dot = networkx.nx_agraph.to_agraph(graph)
if len(dot.edges())> 1000:
print(" skipping")
continue
dot.draw("{}/temp{}{}.pdf".format(outdir, file_label, i), prog="dot")
print("Supported:", supported, "Missing:", missing, "Total breakpoints:", breakpoints)
except:
pass
def visualize_frags(outdir, graphs, options):
from rpy2.robjects import r
utilities.ensure_dir(outdir)
for i, graph in enumerate(graphs):
r.pdf(os.path.join(outdir, "fragments.cluster_{}.pdf".format(i)))
for component in networkx.connected_components(graph):
subgraph = graph.subgraph(component)
ends = [node for node,degree in subgraph.degree_iter() if degree==1]
breakends = [node for node in list(networkx.shortest_simple_paths(subgraph, ends[0], ends[1]))[0]]
# breakends = [breakend_from_label(node) for node in breakends]
breakends = breakends[:-1:2] + breakends[-1:]
# print ")"*100, breakends
for sample, dataset in sorted(options.iter_10xdatasets()):
plot_frags(breakends, options, sample, dataset)
# plot_frags(breakpoints, options, sample, dataset)
r["dev.off"]()
def ensure_dirs(self, to_run):
utilities.ensure_dir(to_run[0].results_dir)
utilities.ensure_dir(to_run[0].working_dir)
utilities.ensure_dir(
os.path.join(self.options.log_dir, to_run[0].__class__.__name__))
def ready_output_dir(options):
utilities.ensure_dir(options.working_dir)
utilities.ensure_dir(options.results_dir)
utilities.ensure_dir(options.log_dir)