def create_rev_com_genome(source_genome_path, write_dir):
source_genome_path = source_genome_path.rstrip('/')
genome_name = source_genome_path.split('/')[-1]
write_dir = os.path.join(write_dir, genome_name)
if not os.path.exists(write_dir):
os.mkdir(write_dir)
source_genome_dict = load_directory(source_genome_path, in_memory=True)
file_shapes = {}
for key in source_genome_dict.keys():
data_arr = source_genome_dict[key].__dict__['_arr'][:].transpose()
##Reverse complement by taking base pair complements as well as reversing
rev_comp_data_arr = data_arr[::-1, ::
-1] #Reverse complement the entrire chromosome
_array_writer['bcolz'](rev_comp_data_arr.astype(np.float32),
os.path.join(write_dir, key))
file_shapes[key] = rev_comp_data_arr.shape
print("Created chromosome %s \n" % (key))
##Write the metadata.json file::
print("Writing metadata file \n")
with open(os.path.join(write_dir, 'metadata.json'), 'w') as fp:
json.dump(
{
'file_shapes': file_shapes,
'type': 'array_{}'.format('bcolz'),
'source': source_genome_path
}, fp)
def create_complemented_genome(source_genome_path, write_dir):
"""
Does not reverse. Only takes complements
"""
source_genome_path = source_genome_path.rstrip('/')
source_genome_dict = load_directory(source_genome_path, in_memory=True)
genome_name = source_genome_path.split('/')[-1]
write_dir = os.path.join(write_dir, genome_name)
if not os.path.exists(write_dir):
os.mkdir(write_dir)
file_shapes = {}
for key in source_genome_dict.keys():
data_arr = source_genome_dict[key].__dict__['_arr'][:].transpose()
##Take the complement by just flipping bases
##The shape is now (4,N)
rev_comp_data_arr = data_arr[::-1]
_array_writer['bcolz'](rev_comp_data_arr.astype(np.float32),
os.path.join(write_dir, key))
file_shapes[key] = rev_comp_data_arr.shape
print("Created chromosome %s \n" % (key))
print("Writing metadata file \n")
##Write the metadata.json file::
with open(os.path.join(write_dir, 'metadata.json'), 'w') as fp:
json.dump(
{
'file_shapes': file_shapes,
'type': 'array_{}'.format('bcolz'),
'source': source_genome_path
}, fp)
frag = args.frag
histone = args.histone
model_path = args.model_path
save_dir = args.save_dir
cuda = args.cuda
os.environ["CUDA_VISIBLE_DEVICES"] = cuda
# Logging directories
srv_dir = os.path.join("/srv", "www", "kundaje", "jesikmin",
"test_experiments", save_dir)
if not os.path.exists(srv_dir):
os.makedirs(srv_dir)
data = Data_Directories()
X_test = load_directory(data.input_atac[day][frag],
in_memory=True)['chr22']._arr
X_test = np.expand_dims(np.nan_to_num(X_test), axis=0)
print "Finished fetching X_test"
print X_test.shape
print "Dimension of ATAC-seq signal (input): {}".format(X_test[0].shape)
y_test = load_directory('/srv/scratch/jesikmin/output/bcolz/',
in_memory=True)['chr22']
y_test = np.expand_dims(y_test, axis=0)
y_test = np.expand_dims(y_test, axis=2)
print "Finished fetching Y_test"
print y_test.shape
print "Dimension of ChIP-seq signal (output): {}".format(y_test[0].shape)
'''
Generator only
'''
help=
"bigwig prefix. Example: a `outfile` prefix results in a `outfile.bw` bigwig file"
)
args = parser.parse_args()
return args
# parse args
args = parse_args()
print(args)
bigwig = '{}.bw'.format(args.output_prefix)
# load data directory
logger.info("Loading genomelake data..")
data = load_directory(args.data_dir, in_memory=True)
file_shapes = {}
for chrom, chrom_data in data.items():
logger.info("Chrom " + str(chrom) + "...")
for _channel_idx in range(5):
channel = np.copy(chrom_data._arr[:, _channel_idx])
output_path = os.path.join(
"/srv/scratch/jesikmin/temp/" + str(_channel_idx), chrom)
os.makedirs(output_path)
_array_writer['bcolz'](channel.astype(np.float32), output_path)
file_shapes[chrom] = (chrom_data._arr.shape[0], )
for idx in range(5):
with open(
os.path.join("/srv/scratch/jesikmin/temp/" + str(idx),
'metadata.json'), 'w') as fpp: