def doWork(ref, chrm, truth, canavar):
ds = prepareDS()
loadChrm(ds, ref, chrm)
i_truth, i_canavar = 1, 2
loadCalls(ds, truth, i_truth, chrm)
loadCalls(ds, canavar, i_canavar, chrm)
computeDistCopyNumbers(ds)
log("MEM: %s" % drdcommon.memory_usage())
log('Done')
def doWork(ref, chrm, truth, canavar):
ds = prepareDS()
loadChrm(ds, ref, chrm)
i_truth, i_canavar = 1, 2
loadCalls(ds, truth, i_truth, chrm)
loadCalls(ds, canavar, i_canavar, chrm)
computeDistCopyNumbers(ds)
log("MEM: %s" % drdcommon.memory_usage())
log('Done')
def __update_counts(self):
n_subs = self.__seen_in_more_than_n_samples(self.subs,
self.at_least_seen)
n_indels = self.__seen_in_more_than_n_samples(self.indels,
self.at_least_seen)
n_genes = self.__seen_in_more_than_n_samples_for_genes(
self.genes_partial, self.at_least_seen)
self.counts[self.n_samples] = (n_subs, n_indels, n_genes)
sys.stderr.write(">> #:%s SUBS:%s INDELS:%s GENES:%s MEM(Mbytes):%s AT_LEAST_SEEN:%s\n" % \
(self.n_samples, n_subs, n_indels, n_genes, drdcommon.memory_usage(), self.at_least_seen))
def __load_species_snp_coordinates(self):
fd = drdcommon.xopen(self.coor_fn)
d = {}
self.d_species_coor = d
n = 0
for l in fd:
n += 1
chrm, coor = l.split()
if not d.has_key(chrm):
d[chrm] = {}
d[chrm][int(coor)] = 1
fd.close()
logging.info("# of coordinates loaded: %d" % n)
logging.info("current memory usage in %dkb" % drdcommon.memory_usage())
def __load_species_snp_coordinates(self):
fd = drdcommon.xopen(self.coor_fn)
d = {}
self.d_species_coor = d
n = 0
for l in fd:
n += 1
chrm, coor = l.split()
if not d.has_key(chrm):
d[chrm] = {}
d[chrm][int(coor)] = 1
fd.close()
logging.info("# of coordinates loaded: %d" % n)
logging.info("current memory usage in %dkb" % drdcommon.memory_usage())
def loadCalls(ds, fn, idx, chrm):
log("Loading calls from %s; idx=%s" % (fn, idx))
chrm_found = False
nbp = 0
for l in drdcommon.xopen(fn):
c, start, end, cnv = l.strip().split()
if c == chrm:
chrm_found = True
for i in range(int(start), int(end)+1):
if nbp % 1000000 == 0:
sys.stderr.write("MEM: %s nbp: %s\r" % (drdcommon.memory_usage(), nbp))
ds[idx][i] = round(float(cnv))
nbp += 1
if not chrm_found:
error("\nCould not find chrm in file. Bailing out.")
log("\n%s bp loaded" % nbp)
def loadCalls(ds, fn, idx, chrm):
log("Loading calls from %s; idx=%s" % (fn, idx))
chrm_found = False
nbp = 0
for l in drdcommon.xopen(fn):
c, start, end, cnv = l.strip().split()
if c == chrm:
chrm_found = True
for i in range(int(start), int(end) + 1):
if nbp % 1000000 == 0:
sys.stderr.write("MEM: %s nbp: %s\r" %
(drdcommon.memory_usage(), nbp))
ds[idx][i] = round(float(cnv))
nbp += 1
if not chrm_found:
error("\nCould not find chrm in file. Bailing out.")
log("\n%s bp loaded" % nbp)
def loadChrm(ds, ref, chrm):
if not os.path.exists(ref):
error("Cannot find reference file: %s", ref)
log("Reading reference genome chrm: %s" % chrm)
i = 1
for l in drdcommon.xopen(ref):
l = l.strip()
if i == 1 and l[0] == '>' and l[1:] == chrm:
continue
if i > 1 and l[0] == '>':
break
for bp in l:
if bp.upper() != 'N':
ds[0][i] = 1
if i % 10000000 == 0:
sys.stderr.write("MEM: %s nbp: %s\r" % (drdcommon.memory_usage(), i))
i += 1
log("\n%s bp read." % i)
def loadChrm(ds, ref, chrm):
if not os.path.exists(ref):
error("Cannot find reference file: %s", ref)
log("Reading reference genome chrm: %s" % chrm)
i = 1
for l in drdcommon.xopen(ref):
l = l.strip()
if i == 1 and l[0] == '>' and l[1:] == chrm:
continue
if i > 1 and l[0] == '>':
break
for bp in l:
if bp.upper() != 'N':
ds[0][i] = 1
if i % 10000000 == 0:
sys.stderr.write("MEM: %s nbp: %s\r" %
(drdcommon.memory_usage(), i))
i += 1
log("\n%s bp read." % i)