def load_dataset(organism):
from ml_data import SequenceNucsData, SimpleHistData
global max_features
global maxlen
print('Load organism: {}'.format(organism))
npath, ppath = './fasta/{}_neg.fa'.format(
organism), './fasta/{}_pos.fa'.format(organism)
print(npath, ppath)
ini = 59
k = 1
max_features = 4**k
samples = SequenceNucsData(npath, ppath, k=k)
samples2 = SimpleHistData(npath, ppath, k=3, upto=True)
X, y = samples.getX(), samples.getY()
X = np.hstack((X, samples2.getX()))
# X = X.reshape(-1, 38, 79, 1).astype('float32')
np.random.seed(int(time.time()))
mask = np.array(np.random.randint(2, size=X.shape[1]) + 1, dtype=bool)
# mask = np.array(np.zeros(X.shape[1]) , dtype=bool)
# mask[59]=1
# mask[60]=1
# mask[61]=1
# mask[49]=1
# mask[50]=1
# mask[51]=1
# mask[24]=1
# mask[25]=1
# mask[26]=1
mask[ini] = 1
print(mask)
X = X[:, mask]
X = X.astype('int32')
# ini = 199
# X = X[:, (ini-30):(ini+11)]
y = y.astype('int32')
print('Input Shapes\nX: {} | y: {}'.format(X.shape, y.shape))
maxlen = X.shape[1]
return X, y
def setup_data(self, npath, ppath):
from ml_data import SimpleHistData
from ml_data import DinucAutoCovarData
self.data_list = []
self.data_list.append(SimpleHistData(npath, ppath, k=4))
self.data_list.append(DinucAutoCovarData(npath, ppath, k=4))
return self.data_list
def setup_data(self, npath, ppath):
from ml_data import SimpleHistData
from ml_data import DinucAutoCovarData
# Join data into a single input vector
data = self.join_data(
SimpleHistData(npath, ppath, k=4),
DinucAutoCovarData(npath, ppath),
)
# Get lenghts of input vectors
lengths = self.get_input_lengths(data)
# Define limits of data type on single vector
limits = self.calc_limits(data)
self.single_input_length = limits[-1][-1]
self.data = data
self.limits = limits
self.lengths = lengths
def setup_data(npath, ppath):
from ml_data import SequenceDinucProperties
from ml_data import SimpleHistData
global data
global limits
global lengths
global shared_input_length
# Join data into a single input vector
data = join_data(SimpleHistData(npath, ppath, k=3))
# Get lenghts of input vectors
lengths = get_input_lengths(data)
# Define limits of data type on single vector
limits = calc_limits(data)
# Compute shared input vector length
shared_input_length = 64
def setup_data(npath, ppath):
from ml_data import SimpleHistData
from ml_data import DinucCrossCovarData
global data
global limits
global lengths
global shared_input_length
# Join data into a single input vector
data = join_data(
SimpleHistData(npath, ppath, k=4, upto=True),
DinucCrossCovarData(npath, ppath, k=3, upto=True),
)
# Get lenghts of input vectors
lengths = get_input_lengths(data)
# Define limits of data type on single vector
limits = calc_limits(data)
# Compute shared input vector length
shared_input_length = limits[-1][-1]
epsilon=0.001,
cooldown=0,
min_lr=0)
#npath = "fasta/Bacillus_non_prom.fa"
#ppath = "fasta/Bacillus_prom.fa"
#npath = "fasta/Arabidopsis_non_prom_big.fa"
#ppath = "fasta/Arabidopsis_non_tata.fa"
npath = "fasta/Ecoli_non_prom.fa"
ppath = "fasta/Ecoli_prom.fa"
#mldata = SequenceNucsData(npath, ppath, k=3)
mldata = SequenceDinucProperties(npath, ppath)
mldata2 = SimpleHistData(npath, ppath, k=4, upto=True)
X = mldata.getX()
X2 = mldata2.getX()
Y = mldata.getY()
print X.shape
print Y.shape
posIndex = numpy.where(Y[:] == 1)[0]
negIndex = numpy.where(Y[:] == 0)[0]
diff = len(negIndex) - len(posIndex)
diff = len(negIndex) - diff
print 'DIFF', diff
patience=10,
verbose=0,
mode='auto')
reduceLR = ReduceLROnPlateau(monitor='val_loss',
factor=0.1,
patience=5,
verbose=0,
mode='auto',
epsilon=0.001,
cooldown=0,
min_lr=0)
npath = "fasta/Bacillus_non_prom.fa"
ppath = "fasta/Bacillus_prom.fa"
#mldata = SequenceNucsData(npath, ppath, k=3)
mldata = SimpleHistData(npath, ppath, k=4)
mldata2 = SequenceNucsData(npath, ppath, k=3)
mldata3 = DinucAutoCovarData(npath, ppath)
X = mldata.getX()
Y = mldata.getY()
X2 = mldata2.getX()
X3 = mldata3.getX()
kf = StratifiedKFold(n_splits=10, shuffle=True, random_state=1234)
kf.get_n_splits(X, Y)
cvscores = []
for train_index, test_index in kf.split(X, Y):