def load_genome(genome_path, res=None, verbose=False):
"""
Search, at a given path, for chromosome folders containing TADbit chromosome
objects saved as files.
:param genome_path: Path where to search for TADbit chomosomes
:param None res: Resolution at which should be saved chromosomes
:param False verbose:
:returns: a dictionary with all TADbit chromosomes found
"""
ref_genome = {}
for crm in listdir(genome_path):
crm_path = os.path.join(genome_path, crm)
if not isdir(crm_path):
continue
for crm_fh in listdir(crm_path):
crm_pik = os.path.join(crm_path, crm_fh)
if not check_pik(crm_pik):
continue
if crm in ref_genome:
raise Exception('More than 1 TADbit chromosome file found\n')
if verbose:
print ' Chromosome:', crm
ref_genome[crm] = load_chromosome(crm_pik)
# check resolution
resolutions = [] if not res else [res]
for crm in ref_genome:
for exp in ref_genome[crm].experiments:
resolutions.append(exp.resolution)
if not all([r == resolutions[0] for r in resolutions]):
raise AssertionError('Not all Experiments have the ' +
'same resolution\n')
res = resolutions[0]
return ref_genome
def load_genome(genome_path, res=None, verbose=False):
"""
Search, at a given path, for chromosome folders containing TADbit chromosome
objects saved as files.
:param genome_path: Path where to search for TADbit chomosomes
:param None res: Resolution at which should be saved chromosomes
:param False verbose:
:returns: a dictionary with all TADbit chromosomes found
"""
ref_genome = {}
for crm in listdir(genome_path):
crm_path = os.path.join(genome_path, crm)
if not isdir(crm_path):
continue
for crm_fh in listdir(crm_path):
crm_pik = os.path.join(crm_path, crm_fh)
if not check_pik(crm_pik):
continue
if crm in ref_genome:
raise Exception('More than 1 TADbit chromosome file found\n')
if verbose:
print ' Chromosome:', crm
ref_genome[crm] = load_chromosome(crm_pik)
# check resolution
resolutions = [] if not res else [res]
for crm in ref_genome:
for exp in ref_genome[crm].experiments:
resolutions.append(exp.resolution)
if not all([r == resolutions[0] for r in resolutions]):
raise AssertionError('Not all Experiments have the ' +
'same resolution\n')
res = resolutions[0]
return ref_genome
def test_05_save_load(self):
test_chr = Chromosome(name='Test Chromosome',
experiment_tads=[exp1, exp2],
experiment_names=['exp1', 'exp2'],
experiment_resolutions=[20000,20000])
test_chr.save_chromosome('lolo', force=True)
test_chr = load_chromosome('lolo')
system('rm -f lolo')
system('rm -f lolo_hic')
def loadData(tdbfile, samples):
my_chrom = load_chromosome(tdbfile)
for i, exp in enumerate(my_chrom.experiments):
try:
label, path = samples[i].split(",")
print(path)
print(exp.name)
exp.load_hic_data(path)
exp.normalize_hic(factor=1)
exp.filter_columns()
except IOError:
print 'file not found for experiment: ' + exp.name
continue
return my_chrom
def test_05_save_load(self):
if CHKTIME:
t0 = time()
test_chr1 = Chromosome(name='Test Chromosome',
experiment_tads=[exp1, exp2],
experiment_names=['exp1', 'exp2'],
experiment_resolutions=[20000, 20000],
silent=True)
test_chr1.save_chromosome('lolo', force=True)
test_chr2 = load_chromosome('lolo')
system('rm -f lolo')
system('rm -f lolo_hic')
self.assertEqual(str(test_chr1.__dict__), str(test_chr2.__dict__))
if CHKTIME:
print '5', time() - t0
def test_05_save_load(self):
if CHKTIME:
t0 = time()
test_chr1 = Chromosome(name='Test Chromosome',
experiment_tads=[exp1, exp2],
experiment_names=['exp1', 'exp2'],
experiment_resolutions=[20000,20000],
silent=True)
test_chr1.save_chromosome('lolo', force=True)
test_chr2 = load_chromosome('lolo')
system('rm -f lolo')
system('rm -f lolo_hic')
self.assertEqual(str(test_chr1.__dict__), str(test_chr2.__dict__))
if CHKTIME:
print '5', time() - t0
def test_05_save_load(self):
if ONLY and not "05" in ONLY:
return
if CHKTIME:
t0 = time()
test_chr1 = Chromosome(name="Test Chromosome",
experiment_tads=[exp1, exp2],
experiment_names=["exp1", "exp2"],
experiment_resolutions=[20000, 20000],
silent=True)
test_chr1.save_chromosome("lolo", force=True)
test_chr2 = load_chromosome("lolo")
system("rm -f lolo")
system("rm -f lolo_hic")
self.assertEqual(str(test_chr1.__dict__), str(test_chr2.__dict__))
if CHKTIME:
print "5", time() - t0
def test_05_save_load(self):
if ONLY and ONLY != "05":
return
if CHKTIME:
t0 = time()
test_chr1 = Chromosome(
name="Test Chromosome",
experiment_tads=[exp1, exp2],
experiment_names=["exp1", "exp2"],
experiment_resolutions=[20000, 20000],
silent=True,
)
test_chr1.save_chromosome("lolo", force=True)
test_chr2 = load_chromosome("lolo")
system("rm -f lolo")
system("rm -f lolo_hic")
self.assertEqual(str(test_chr1.__dict__), str(test_chr2.__dict__))
if CHKTIME:
print "5", time() - t0
def main():
"""
main function
"""
opts = get_options()
res = opts.res
if opts.genomes:
# load all chromosomes of reference genomes
ref_genome = {}
for crm in listdir(opts.ref_genome):
crm_path = opts.ref_genome + crm + '/'
if not isdir(crm_path):
continue
for crm_fh in listdir(crm_path):
crm_pik = crm_path + crm_fh
if not check_pik(crm_pik):
continue
ref_genome[crm] = load_chromosome(crm_pik)
if not opts.res:
resolutions = []
for crm in ref_genome:
for exp in ref_genome[crm].experiments:
resolutions.append(exp.resolution)
if not all([r == resolutions[0] for r in resolutions]):
raise AssertionError('Not all Experiments have the ' +
'same resolution\n')
res = resolutions[0]
alt_genomes = {}
for i, genome in enumerate(opts.genomes):
alt_genomes[i] = {}
for crm in listdir(genome):
crm_path = genome + crm + '/'
if not isdir(crm_path):
continue
for crm_fh in listdir(crm_path):
crm_pik = crm_path + crm_fh
if not check_pik(crm_pik):
continue
try:
alt_genomes[i][crm] = load_chromosome(crm_pik)
except:
print('SKIPPING: {} \n not a valid ' +
'chromosome').format(crm_pik)
genome = {}
for crm in alt_genomes[i]:
genome = remap_chr(alt_genomes[i][crm],
crm,
'/tmp/',
opts.lft_path,
opts.chain_path,
genome=genome)
reorder(genome)
for exp in genome:
for crm in genome[exp]:
try:
ref_genome[crm].add_experiment(
exp, res, tad_handler=genome[exp][crm])
except KeyError:
print('Chromosome {} skipped, not in reference ' +
'genome').format(crm)
system('mkdir -p ' + opts.out_path)
for crm in ref_genome:
system('mkdir -p ' + opts.out_path + '/' + crm)
out_f = opts.out_path + '/' + crm + '/chr' + crm + '.tdb'
ref_genome[crm].save_chromosome(out_f, force=True)
def main():
"""
main function
"""
opts = get_options()
res = opts.res
if opts.genomes:
# load all chromosomes of reference genomes
ref_genome = {}
for crm in listdir(opts.ref_genome):
crm_path = opts.ref_genome + crm + '/'
if not isdir(crm_path):
continue
for crm_fh in listdir(crm_path):
crm_pik = crm_path + crm_fh
if not check_pik(crm_pik):
continue
ref_genome[crm] = load_chromosome(crm_pik)
if not opts.res:
resolutions = []
for crm in ref_genome:
for exp in ref_genome[crm].experiments:
resolutions.append(exp.resolution)
if not all([r == resolutions[0] for r in resolutions]):
raise AssertionError('Not all Experiments have the ' +
'same resolution\n')
res = resolutions[0]
alt_genomes = {}
for i, genome in enumerate(opts.genomes):
alt_genomes[i] = {}
for crm in listdir(genome):
crm_path = genome + crm + '/'
if not isdir(crm_path):
continue
for crm_fh in listdir(crm_path):
crm_pik = crm_path + crm_fh
if not check_pik(crm_pik):
continue
try:
alt_genomes[i][crm] = load_chromosome(crm_pik)
except:
print ('SKIPPING: {} \n not a valid ' +
'chromosome').format(crm_pik)
genome = {}
for crm in alt_genomes[i]:
genome = remap_chr(alt_genomes[i][crm], crm, '/tmp/',
opts.lft_path, opts.chain_path,
genome=genome)
reorder(genome)
for exp in genome:
for crm in genome[exp]:
try:
ref_genome[crm].add_experiment(
exp, res, tad_handler=genome[exp][crm])
except KeyError:
print ('Chromosome {} skipped, not in reference ' +
'genome').format(crm)
system('mkdir -p ' + opts.out_path)
for crm in ref_genome:
system('mkdir -p ' + opts.out_path + '/' + crm)
out_f = opts.out_path + '/' + crm + '/chr' + crm + '.tdb'
ref_genome[crm].save_chromosome(out_f, force=True)
