def gen_clusters_from_data(fmeth, min_clust_size=3, max_clust_size=200, n=20, w=0.5):
header = nopen(fmeth).readline().rstrip().split("\t")[1:]
clust_gen = (c for c in aclust(feature.feature_gen(fmeth),
max_dist=500,
max_skip=1,
linkage=0.5)
if len(c) >= min_clust_size
and len(c) <= max_clust_size)
for cluster in generate(clust_gen, header, n=n, w=w):
yield cluster
def gen_clusters_from_data(fmeth,
min_clust_size=3,
max_clust_size=200,
n=20,
w=0.5):
header = nopen(fmeth).readline().rstrip().split("\t")[1:]
clust_gen = (c for c in aclust(
feature.feature_gen(fmeth), max_dist=500, max_skip=1, linkage=0.5)
if len(c) >= min_clust_size and len(c) <= max_clust_size)
for cluster in generate(clust_gen, header, n=n, w=w):
yield cluster
def main(argv=sys.argv[1:]):
import argparse
ap = argparse.ArgumentParser()
ap.add_argument("--n-samples", type=int, default=20,
help="number of samples from each group to simulate")
ap.add_argument("-w", type=float, default=0,
help="weight parameter 0 is random, 1 is strong separation"
" between simulated groups")
ap.add_argument("methylation",
help="input methylation data from which to simulate data")
ap.add_argument("prefix",
help="output prefix. prefix.meth.txt, prefix.covs.txt "
"and prefix.clusters.txt" " be created")
cp = ap.add_argument_group('clustering parameters')
cp.add_argument('--rho-min', type=float, default=0.3,
help="minimum correlation to merge 2 probes")
cp.add_argument('--min-cluster-size', type=int, default=2,
help="minimum cluster size on which to run model: "
"must be at least 2")
cp.add_argument('--linkage', choices=['single', 'complete'],
default='complete', help="linkage method")
cp.add_argument('--max-dist', default=500, type=int,
help="maximum distance beyond which a probe can not be"
" added to a cluster")
args = ap.parse_args(argv)
samples = np.array(nopen(args.methylation).readline().rstrip().split("\t")[1:])
cluster_gen = (c for c in aclust(feature.feature_gen(args.methylation,
rho_min=args.rho_min),
max_dist=args.max_dist,
max_skip=1,
linkage=args.linkage))
def check_cluster(clust):
return len(clust) >= args.min_cluster_size
n, w = args.n_samples, args.w
fh_meth = nopen("%s.meth.txt" % args.prefix, "w")
fh_clst = nopen("%s.clusters.txt" % args.prefix, "w")
with nopen("%s.covs.txt" % args.prefix, "w") as fh_covs:
print >>fh_covs, "id\tcase"
print >>fh_covs, "\n".join(
["case_%i\t1" % i for i in range(n)] +
["ctrl_%i\t0" % i for i in range(n)])
print >>fh_meth, "probe\t" + "\t".join(
["case_%i" % i for i in range(n)] +
["ctrl_%i" % i for i in range(n)])
print >>fh_clst, "chrom\tstart\tend\tw\tprobes"
for i, ocluster in enumerate(cluster_gen):
is_cluster = check_cluster(ocluster)
cluster = simulate_cluster(ocluster, samples, n, w if is_cluster else 0)
cluster.to_csv(fh_meth, sep="\t", index=True, header=False,
float_format="%.3f")
chrom, start, end = ocluster[0].group, ocluster[0].start, ocluster[-1].end
probes = ",".join("%s:%i" % (o.group, o.end) for o in ocluster)
if is_cluster:
print >>fh_clst, "{chrom}\t{start}\t{end}\t{w}\t{probes}".format(**locals())
print >>sys.stderr, "wrote:", fh_meth.name
def main(argv=sys.argv[1:]):
import argparse
ap = argparse.ArgumentParser()
ap.add_argument("--n-samples",
type=int,
default=20,
help="number of samples from each group to simulate")
ap.add_argument("-w",
type=float,
default=0,
help="weight parameter 0 is random, 1 is strong separation"
" between simulated groups")
ap.add_argument("methylation",
help="input methylation data from which to simulate data")
ap.add_argument("prefix",
help="output prefix. prefix.meth.txt, prefix.covs.txt "
"and prefix.clusters.txt"
" be created")
cp = ap.add_argument_group('clustering parameters')
cp.add_argument('--rho-min',
type=float,
default=0.3,
help="minimum correlation to merge 2 probes")
cp.add_argument('--min-cluster-size',
type=int,
default=2,
help="minimum cluster size on which to run model: "
"must be at least 2")
cp.add_argument('--linkage',
choices=['single', 'complete'],
default='complete',
help="linkage method")
cp.add_argument('--max-dist',
default=500,
type=int,
help="maximum distance beyond which a probe can not be"
" added to a cluster")
args = ap.parse_args(argv)
samples = np.array(
nopen(args.methylation).readline().rstrip().split("\t")[1:])
cluster_gen = (c for c in aclust(feature.feature_gen(args.methylation,
rho_min=args.rho_min),
max_dist=args.max_dist,
max_skip=1,
linkage=args.linkage))
def check_cluster(clust):
return len(clust) >= args.min_cluster_size
n, w = args.n_samples, args.w
fh_meth = nopen("%s.meth.txt" % args.prefix, "w")
fh_clst = nopen("%s.clusters.txt" % args.prefix, "w")
with nopen("%s.covs.txt" % args.prefix, "w") as fh_covs:
print >> fh_covs, "id\tcase"
print >> fh_covs, "\n".join(["case_%i\t1" % i for i in range(n)] +
["ctrl_%i\t0" % i for i in range(n)])
print >> fh_meth, "probe\t" + "\t".join(["case_%i" % i for i in range(n)] +
["ctrl_%i" % i for i in range(n)])
print >> fh_clst, "chrom\tstart\tend\tw\tprobes"
for i, ocluster in enumerate(cluster_gen):
is_cluster = check_cluster(ocluster)
cluster = simulate_cluster(ocluster, samples, n,
w if is_cluster else 0)
cluster.to_csv(fh_meth,
sep="\t",
index=True,
header=False,
float_format="%.3f")
chrom, start, end = ocluster[0].group, ocluster[0].start, ocluster[
-1].end
probes = ",".join("%s:%i" % (o.group, o.end) for o in ocluster)
if is_cluster:
print >> fh_clst, "{chrom}\t{start}\t{end}\t{w}\t{probes}".format(
**locals())
print >> sys.stderr, "wrote:", fh_meth.name