def load_data_mongoloids_hannum_cpgs():
from mongoloids_cpg_hannum_config import config
start = timeit.default_timer()
X = np.genfromtxt(config.ifname("hannum_cpgs_beta"),
dtype='float32',
delimiter=' ')[:, 1:]
cpgs_names = np.genfromtxt(config.ifname("hannum_cpgs_beta"),
dtype='str',
usecols=0)
config.params["num_cpgs"].value = min(cpgs_names.size,
config.params["num_cpgs"].value)
stop = timeit.default_timer()
print('Data loaded: ', stop - start)
print(X.dtype, X.shape)
sys.stdout.flush()
X = X.T
y = np.zeros((X.shape[0], 1), dtype='uint8')
y[config.params["mongoloids_mask"].value] = 0
y[config.params["siblings_mask"].value] = 1
y[config.params["mothers_mask"].value] = 2
print(X.shape, config.params["num_cpgs"].value)
sys.stdout.flush()
mask = config.params[config.params["kde_mask"].value].value
return X, y, X[mask, :], cpgs_names
def load_data_down_GSE52588():
from configurations.config_down_GSE52588 import config
start = timeit.default_timer()
print(config.ifname("x"))
X = np.genfromtxt(config.ifname("x"), dtype='float32', delimiter=' ')[:,
1:]
genes_names = np.genfromtxt(config.ifname("x"), dtype='str', usecols=0)
config.params["num_genes"].value = min(genes_names.size,
config.params["num_genes"].value)
genes_dict = dict((v, i) for i, v in enumerate(genes_names))
#ranged_genes = np.genfromtxt(config.ifname("ranged_genes"), dtype='str', usecols = 0)
stop = timeit.default_timer()
print('Data loaded: ', stop - start)
print(X.dtype, X.shape)
sys.stdout.flush()
#X = np.random.rand(len(genes_names), 656)
#genes_names = ranged_genes[:config.params["num_genes"].value]
genes_names = genes_names[:config.params["num_genes"].value]
indices = np.array([genes_dict[x] for x in genes_names])
X = X[indices, :].T
#X = X.T
y, mask = get_classes(config, X)
print(X.shape, config.params["num_genes"].value)
sys.stdout.flush()
return X, y, mask, genes_names
def load_data_mongoloids_cpgs():
from mongoloids_cpgs_config import config
start = timeit.default_timer()
X = np.genfromtxt(config.ifname("cpgs"), dtype='float32',
delimiter=' ')[:, 2:]
genes_names = np.genfromtxt(config.ifname("cpgs"), dtype='str', usecols=0)
cpgs_names = np.genfromtxt(config.ifname("cpgs"), dtype='str', usecols=1)
config.params["num_cpgs"].value = min(cpgs_names.size,
config.params["num_cpgs"].value)
genes_dict = dict((v, i) for i, v in enumerate(genes_names))
#ranged_genes = np.genfromtxt(config.ifname("ranged_genes"), dtype='str', usecols = 0)
stop = timeit.default_timer()
print('Data loaded: ', stop - start)
print(X.dtype, X.shape)
sys.stdout.flush()
#X = np.random.rand(len(genes_names), 656)
#genes_names = ranged_genes[:config.params["num_genes"].value]
#genes_names = genes_names[:config.params["num_genes"].value]
#indices = np.array([genes_dict[x] for x in genes_names])
#X = X[indices, :].T
X = X.T
#X = X.T
y = np.zeros((X.shape[0], 1), dtype='uint8')
y[config.params["mongoloids_mask"].value] = 0
y[config.params["siblings_mask"].value] = 1
y[config.params["mothers_mask"].value] = 2
print(X.shape, config.params["num_cpgs"].value)
sys.stdout.flush()
mask = config.params[config.params["kde_mask"].value].value
return X, y, X[mask, :], genes_names, cpgs_names
def load_data_down_GSE52588_cpgs(is_small=False):
from configurations.config_down_GSE52588_cpg import config
start = timeit.default_timer()
import pandas as pd
X = np.genfromtxt(config.ifname("x"),
dtype='float32',
delimiter='\t',
skip_header=0)[:, 1:]
cpgs_names = np.genfromtxt(config.ifname("x"),
dtype='str',
skip_header=0,
usecols=0)
config.params["num_cpgs"].value = min(cpgs_names.size,
config.params["num_cpgs"].value)
stop = timeit.default_timer()
print('Data loaded: ', stop - start)
print(X.dtype, X.shape)
sys.stdout.flush()
if is_small:
from annotations.cpgs import cpgs_annotation
cpgs = cpgs_annotation(config.ifname('cpgs_annotations'))
bad_cpgs = np.loadtxt(config.ifname('bad_cpgs'), dtype='str')
subset_cpg_names, ids = cpgs.get_cpgs({
'gene_out': [np.NaN],
'cpgs_in': cpgs_names,
'chr_out': ['X', 'Y'],
'geotype_in': ['Island'],
'cpgs_out': bad_cpgs
})
cpgs_names, indices, _ = np.intersect1d(cpgs_names,
subset_cpg_names,
return_indices=True)
#cpgs_island = np.array(cpgs_island)
#cpgs_names = np.array(cpgs_names)
#_, indices, _ = np.intersect1d(cpgs_names, cpgs_all, return_indices = True)
#cpgs_names = cpgs_names[indices]
X = X[indices, :]
X = X.T
good_cpgs = ~np.isnan(X).any(axis=0)
X = X[:, good_cpgs]
cpgs_names = cpgs_names[good_cpgs]
print(X.shape)
#X = X.T
config.params["num_cpgs"].value = min(X.shape[1],
config.params["num_cpgs"].value)
y, mask = get_classes(config, X)
#print y
#print mask
print(X.shape, config.params["num_cpgs"].value, y.shape, cpgs_names.shape)
sys.stdout.flush()
return X, y, mask, cpgs_names