from keras.layers import Dense, Dropout, Activation, Flatten
from keras.layers import Convolution2D, MaxPooling2D
from keras.utils import np_utils
from keras.datasets import mnist
#############################################
seed = 123 # for reproducibility
np.random.seed(seed)
# 1. Load data into train and test sets
X, y = load_data("../data/splice.data.txt") # sequences, labels
X = get_rep_mats(X) # convert to array of representation matrices
for i in X: # CUSTOM reshape
for idx, j in enumerate(i):
i[idx] = j[0]
y = conv_labels(y) # convert to integer labels
X = np.asarray(X) # work with np arrays
Y = np.asarray(y)
# define 10-fold cross validation test harness
kfold = StratifiedKFold(n_splits=10, shuffle=True, random_state=seed)
scores = []
for train, test in kfold.split(X, Y):
print "====> FOLD [" + str(len(scores) + 1) + "]"
# 2. Preprocess input data
X_train = X[train].reshape(X[train].shape[0], 1, 58, 64)
X_test = X[test].reshape(X[test].shape[0], 1, 58, 64)
X_train = X_train.astype('float32')
X_test = X_test.astype('float32')
# 1. Load data into train and test sets
X, y = load_data("../data/promoters.data.txt") # sequences, labels
X = get_rep_mats(X) # convert to array of representation matrices
############
# In[ ]:
for i in X:
for idx, j in enumerate(i):
i[idx] = j[0]
############
# In[ ]:
y = conv_labels(y, "promoter") # convert to integer labels
X = np.asarray(X) # work with np arrays
y = np.asarray(y)
X_train = X[0:90]
X_test = X[90:]
y_train = y[0:90]
y_test = y[90:]
# In[ ]:
# 2. Preprocess input data
X_train = X_train.reshape(X_train.shape[0], 1, 55,
64) # (90, 55, 64) --> (90, 1, 55, 64)
X_test = X_test.reshape(X_test.shape[0], 1, 55, 64)
X_train = X_train.astype('float32')
X_test = X_test.astype('float32')
seq = seq.upper() # b/c rep matrix built on uppercase
seq = seq.replace("\t","") # present in promoter
seq = seq.replace("N","A") # undetermined nucleotides in splice
seq = seq.replace("D","G")
seq = seq.replace("S","C")
seq = seq.replace("R","G")
#####
labels.append(label)
seqs.append(seq)
f.close()
return seqs, labels
# In[11]:
if __name__ == "__main__":
# reading in splice junction input data and converting to required format
seqs, labels = load_data("../data/splice.data.txt")
lbls_mod = conv_labels(labels)
seqs_mod = get_rep_mats(seqs)
print (len(seqs_mod))
print (len(lbls_mod))
# In[ ]: