seyyedsalehi
/
PFP-WGAN
Watch 1
Star 0
Fork 0
Code Issues 0 Pull Requests 0 Releases 0 Wiki Activity
Browse Source

Add Source files

master
seydsalehi 5 years ago
parent
32e89e747c
commit
0b25badeab
6 changed files with 2021 additions and 0 deletions
Unified View Show Diff Stats
  1. 128
    0
      annotator.py
  2. 191
    0
      conditional_wgan_wrapper_exp1.py
  3. 193
    0
      conditional_wgan_wrapper_exp2.py
  4. 868
    0
      pfp_conditional_wgan_exp1.py
  5. 636
    0
      pfp_conditional_wgan_exp2.py
  6. 5
    0
      utils.py

+ 128
- 0
annotator.py View File

#!/usr/bin/env python
import click as ck
import numpy as np
import pandas as pd
from keras.models import load_model
from aaindex import INVALID_ACIDS
MAXLEN = 1002
@ck.command()
@ck.option('--in-file', '-i', help='Input FASTA file', required=True)
@ck.option('--threshold', '-t', default=0.3, help='Prediction threshold')
@ck.option('--batch-size', '-bs', default=1, help='Batch size for prediction model')
@ck.option('--include-long-seq', '-ils', is_flag=True, help='Include long sequences')
@ck.option('--onto', '-go', default='bp', help='Sub-ontology to be predicted: bp, mf, cc')
def main(in_file, threshold, batch_size, include_long_seq, onto):
out_file = onto +'_results.txt'
chunk_size = 1000
global model
global functions
ngram_df = pd.read_pickle('data/models/ngrams.pkl')
global vocab
vocab = {}
global gram_len
for key, gram in enumerate(ngram_df['ngrams']):
vocab[gram] = key + 1
gram_len = len(ngram_df['ngrams'][0])
print(('Gram length:', gram_len))
print(('Vocabulary size:', len(vocab)))
model = load_model('Data/data1/models/model_%s.h5' % onto)
df = pd.read_pickle('Data/data1/models/%s.pkl' % onto)
functions = df['functions']
w = open(out_file, 'w')
for ids, sequences in read_fasta(in_file, chunk_size, include_long_seq):
data = get_data(sequences)
results = predict(data, model, threshold, batch_size)
for i in range(len(ids)):
w.write(ids[i])
w.write('\n')
for res in results[i]:
w.write(res)
w.write('\n')
w.close()
def is_ok(seq):
for c in seq:
if c in INVALID_ACIDS:
return False
return True
def read_fasta(filename, chunk_size, include_long_seq):
seqs = list()
info = list()
seq = ''
inf = ''
with open(filename) as f:
for line in f:
line = line.strip()
if line.startswith('>'):
if seq != '':
if is_ok(seq):
if include_long_seq:
seqs.append(seq)
info.append(inf)
if len(info) == chunk_size:
yield (info, seqs)
seqs = list()
info = list()
elif len(seq) <= MAXLEN:
seqs.append(seq)
info.append(inf)
if len(info) == chunk_size:
yield (info, seqs)
seqs = list()
info = list()
else:
print(('Ignoring sequence {} because its length > 1002'
.format(inf)))
else:
print(('Ignoring sequence {} because of ambigious AA'
.format(inf)))
seq = ''
inf = line[1:].split()[0]
else:
seq += line
seqs.append(seq)
info.append(inf)
yield (info, seqs)
def get_data(sequences):
n = len(sequences)
data = np.zeros((n, 1000), dtype=np.float32)
for i in range(len(sequences)):
seq = sequences[i]
for j in range(min(MAXLEN, len(seq)) - gram_len + 1):
data[i, j] = vocab[seq[j: (j + gram_len)]]
return data
def predict(data, model, threshold, batch_size):
n = data.shape[0]
result = list()
for i in range(n):
result.append(list())
predictions = model.predict(data, batch_size=batch_size, verbose=1)
for i in range(n):
pred = (predictions[i] >= threshold).astype('int32')
for j in range(len(functions)):
if pred[j] == 1:
result[i].append(functions[j] + ' with score ' + '%.2f' % predictions[i][j])
return result
if __name__ == '__main__':
main()

+ 191
- 0
conditional_wgan_wrapper_exp1.py View File

from functools import partial

import numpy as np
from keras import backend as K
from keras.layers import Input, RepeatVector, Multiply, Lambda, Permute
from keras.layers.merge import _Merge
from keras.models import Model
from keras.optimizers import Adam, SGD
from keras.losses import binary_crossentropy

def wasserstein_loss(y_true, y_pred):
return K.mean(-y_true * y_pred)



def generator_recunstruction_loss(y_true, y_pred, enableTrain):

def rescale_back_labels(labels):
rescaled_back_labels = (labels / 2) + 0.5
return rescaled_back_labels

return binary_crossentropy(rescale_back_labels(y_true), rescale_back_labels(y_pred)) #* enableTrain


global enable_train
enable_train = Input(shape = (1,))
global generator_recunstruction_loss_new
generator_recunstruction_loss_new = partial(generator_recunstruction_loss, enableTrain = enable_train)
generator_recunstruction_loss_new.__name__ = 'generator_recunstruction_loss_new'


def gradient_penalty_loss(y_true, y_pred, averaged_samples, gradient_penalty_weight):
"""Calculates the gradient penalty loss for a batch of "averaged" samples.

In Improved WGANs, the 1-Lipschitz constraint is enforced by adding a term to the loss function
that penalizes the network if the gradient norm moves away from 1. However, it is impossible to evaluate
this function at all points in the input space. The compromise used in the paper is to choose random points
on the lines between real and generated samples, and check the gradients at these points. Note that it is the
gradient w.r.t. the input averaged samples, not the weights of the discriminator, that we're penalizing!

In order to evaluate the gradients, we must first run samples through the generator and evaluate the loss.
Then we get the gradients of the discriminator w.r.t. the input averaged samples.
The l2 norm and penalty can then be calculated for this gradient.

Note that this loss function requires the original averaged samples as input, but Keras only supports passing
y_true and y_pred to loss functions. To get around this, we make a partial() of the function with the
averaged_samples argument, and use that for model training."""
# first get the gradients:
# assuming: - that y_pred has dimensions (batch_size, 1)
# - averaged_samples has dimensions (batch_size, nbr_features)
# gradients afterwards has dimension (batch_size, nbr_features), basically
# a list of nbr_features-dimensional gradient vectors
gradients = K.gradients(y_pred, averaged_samples)[0]
# compute the euclidean norm by squaring ...
gradients_sqr = K.square(gradients)
# ... summing over the rows ...
gradients_sqr_sum = K.sum(gradients_sqr,
axis=np.arange(1, len(gradients_sqr.shape)))
# ... and sqrt
gradient_l2_norm = K.sqrt(gradients_sqr_sum)
# compute lambda * (1 - ||grad||)^2 still for each single sample
gradient_penalty = gradient_penalty_weight * K.square(1 - gradient_l2_norm)
# return the mean as loss over all the batch samples
return K.mean(gradient_penalty)

def WGAN_wrapper(generator, discriminator, generator_input_shape, discriminator_input_shape, discriminator_input_shape2,
batch_size, gradient_penalty_weight, embeddings):

BATCH_SIZE = batch_size
GRADIENT_PENALTY_WEIGHT = gradient_penalty_weight


def set_trainable_state(model, state):
for layer in model.layers:
layer.trainable = state
model.trainable = state


class RandomWeightedAverage(_Merge):
"""Takes a randomly-weighted average of two tensors. In geometric terms, this outputs a random point on the line
between each pair of input points.

Inheriting from _Merge is a little messy but it was the quickest solution I could think of.
Improvements appreciated."""

def _merge_function(self, inputs):
weights = K.random_uniform((BATCH_SIZE, 1))
print(inputs[0])
return (weights * inputs[0]) + ((1 - weights) * inputs[1])

# The generator_model is used when we want to train the generator layers.
# As such, we ensure that the discriminator layers are not trainable.
# Note that once we compile this model, updating .trainable will have no effect within it. As such, it
# won't cause problems if we later set discriminator.trainable = True for the discriminator_model, as long
# as we compile the generator_model first.


def mul(input):
embedds = K.variable(np.expand_dims(embeddings.T, 0), dtype='float32')
batch_size = K.shape(input)[0]
expanded_embed = K.tile(embedds, (batch_size, 1, 1))
expanded_input = RepeatVector(100)(input)
masked = Multiply()([expanded_embed, expanded_input])
return K.permute_dimensions(masked, [0,2,1])


set_trainable_state(discriminator, False)
set_trainable_state(generator, True)

#enable_train = Input(shape = (1,))
generator_input = Input(shape=generator_input_shape)
generator_layers = generator(generator_input)

#masked_generator_layers = Lambda(mul)(generator_layers)

#discriminator_noise_in = Input(shape=(1,))
#input_seq_g = Input(shape = discriminator_input_shape2)
discriminator_layers_for_generator= discriminator([generator_layers, generator_input])
generator_model = Model(inputs=[generator_input, enable_train],
outputs=[generator_layers, discriminator_layers_for_generator])
# We use the Adam paramaters from Gulrajani et al.
#global generator_recunstruction_loss_new
#generator_recunstruction_loss_new = partial(generator_recunstruction_loss, enableTrain = enable_train)
#generator_recunstruction_loss_new.__name__ = 'generator_RLN'

loss = [generator_recunstruction_loss_new, wasserstein_loss]
loss_weights = [10000, 1]
generator_model.compile(optimizer=Adam(lr=1E-4, beta_1=0.9, beta_2=0.999, epsilon=1e-08),
loss=loss, loss_weights=loss_weights)

# Now that the generator_model is compiled, we can make the discriminator layers trainable.

set_trainable_state(discriminator, True)
set_trainable_state(generator, False)

# The discriminator_model is more complex. It takes both real image samples and random noise seeds as input.
# The noise seed is run through the generator model to get generated images. Both real and generated images
# are then run through the discriminator. Although we could concatenate the real and generated images into a
# single tensor, we don't (see model compilation for why).
real_samples = Input(shape=discriminator_input_shape)
input_seq = Input(shape = discriminator_input_shape2)
generator_input_for_discriminator = Input(shape=generator_input_shape)
generated_samples_for_discriminator = generator(generator_input_for_discriminator)
#masked_real_samples = Lambda(mul)(real_samples)
#masked_generated_samples_for_discriminator= Lambda(mul)(generated_samples_for_discriminator)
discriminator_output_from_generator = discriminator([generated_samples_for_discriminator, generator_input_for_discriminator])
discriminator_output_from_real_samples= discriminator([real_samples, input_seq])

# We also need to generate weighted-averages of real and generated samples, to use for the gradient norm penalty.
averaged_samples = RandomWeightedAverage()([real_samples, generated_samples_for_discriminator])
average_seq = RandomWeightedAverage()([input_seq, generator_input_for_discriminator])
# We then run these samples through the discriminator as well. Note that we never really use the discriminator
# output for these samples - we're only running them to get the gradient norm for the gradient penalty loss.
#print('hehehe')
#print(averaged_samples)
#masked_averaged_samples = Lambda(mul)(averaged_samples)
averaged_samples_out = discriminator([averaged_samples, average_seq])

# The gradient penalty loss function requires the input averaged samples to get gradients. However,
# Keras loss functions can only have two arguments, y_true and y_pred. We get around this by making a partial()
# of the function with the averaged samples here.

partial_gp_loss = partial(gradient_penalty_loss,
averaged_samples=averaged_samples,
gradient_penalty_weight=GRADIENT_PENALTY_WEIGHT)
partial_gp_loss.__name__ = 'gradient_penalty' # Functions need names or Keras will throw an error

# Keras requires that inputs and outputs have the same number of samples. This is why we didn't concatenate the
# real samples and generated samples before passing them to the discriminator: If we had, it would create an
# output with 2 * BATCH_SIZE samples, while the output of the "averaged" samples for gradient penalty
# would have only BATCH_SIZE samples.

# If we don't concatenate the real and generated samples, however, we get three outputs: One of the generated
# samples, one of the real samples, and one of the averaged samples, all of size BATCH_SIZE. This works neatly!
discriminator_model = Model(inputs=[real_samples, generator_input_for_discriminator, input_seq],
outputs=[discriminator_output_from_real_samples,
discriminator_output_from_generator,
averaged_samples_out])
loss_weights2 = [1, 1, 1]
# We use the Adam paramaters from Gulrajani et al. We use the Wasserstein loss for both the real and generated
# samples, and the gradient penalty loss for the averaged samples.
discriminator_model.compile(optimizer=SGD(clipvalue=0.01), # Adam(lr=4E-4, beta_1=0.9, beta_2=0.999, epsilon=1e-08),
loss=[wasserstein_loss,
wasserstein_loss,
partial_gp_loss], loss_weights = loss_weights2)

# set_trainable_state(discriminator, True)
# set_trainable_state(generator, True)

return generator_model, discriminator_model

+ 193
- 0
conditional_wgan_wrapper_exp2.py View File

from functools import partial
import numpy as np
from keras import backend as K
from keras.layers import Input, RepeatVector, Multiply, Lambda, Permute
from keras.layers.merge import _Merge
from keras.models import Model
from keras.optimizers import Adam, SGD
from keras.losses import binary_crossentropy
def wasserstein_loss(y_true, y_pred):
"""Calculates the Wasserstein loss for a sample batch.
The Wasserstein loss function is very simple to calculate. In a standard GAN, the discriminator
has a sigmoid output, representing the probability that samples are real or generated. In Wasserstein
GANs, however, the output is linear with no activation function! Instead of being constrained to [0, 1],
the discriminator wants to make the distance between its output for real and generated samples as large as possible.
The most natural way to achieve this is to label generated samples -1 and real samples 1, instead of the
0 and 1 used in normal GANs, so that multiplying the outputs by the labels will give you the loss immediately.
Note that the nature of this loss means that it can be (and frequently will be) less than 0."""
return K.mean(-y_true * y_pred)
def generator_recunstruction_loss(y_true, y_pred, enableTrain):
def rescale_back_labels(labels):
rescaled_back_labels = (labels / 2) + 0.5
return rescaled_back_labels
return binary_crossentropy(rescale_back_labels(y_true), rescale_back_labels(y_pred)) # * enableTrain
global enable_train
enable_train = Input(shape=(1,))
global generator_recunstruction_loss_new
generator_recunstruction_loss_new = partial(generator_recunstruction_loss, enableTrain=enable_train)
generator_recunstruction_loss_new.__name__ = 'generator_recunstruction_loss_new'
def gradient_penalty_loss(y_true, y_pred, averaged_samples, gradient_penalty_weight):
"""Calculates the gradient penalty loss for a batch of "averaged" samples.
In Improved WGANs, the 1-Lipschitz constraint is enforced by adding a term to the loss function
that penalizes the network if the gradient norm moves away from 1. However, it is impossible to evaluate
this function at all points in the input space. The compromise used in the paper is to choose random points
on the lines between real and generated samples, and check the gradients at these points. Note that it is the
gradient w.r.t. the input averaged samples, not the weights of the discriminator, that we're penalizing!
In order to evaluate the gradients, we must first run samples through the generator and evaluate the loss.
Then we get the gradients of the discriminator w.r.t. the input averaged samples.
The l2 norm and penalty can then be calculated for this gradient.
Note that this loss function requires the original averaged samples as input, but Keras only supports passing
y_true and y_pred to loss functions. To get around this, we make a partial() of the function with the
averaged_samples argument, and use that for model training."""
# first get the gradients:
# assuming: - that y_pred has dimensions (batch_size, 1)
# - averaged_samples has dimensions (batch_size, nbr_features)
# gradients afterwards has dimension (batch_size, nbr_features), basically
# a list of nbr_features-dimensional gradient vectors
gradients = K.gradients(y_pred, averaged_samples)[0]
# compute the euclidean norm by squaring ...
gradients_sqr = K.square(gradients)
# ... summing over the rows ...
gradients_sqr_sum = K.sum(gradients_sqr,
axis=np.arange(1, len(gradients_sqr.shape)))
# ... and sqrt
gradient_l2_norm = K.sqrt(gradients_sqr_sum)
# compute lambda * (1 - ||grad||)^2 still for each single sample
gradient_penalty = gradient_penalty_weight * K.square(1 - gradient_l2_norm)
# return the mean as loss over all the batch samples
return K.mean(gradient_penalty)
def WGAN_wrapper(generator, discriminator, generator_input_shape, discriminator_input_shape, discriminator_input_shape2,
batch_size, gradient_penalty_weight):
BATCH_SIZE = batch_size
GRADIENT_PENALTY_WEIGHT = gradient_penalty_weight
def set_trainable_state(model, state):
for layer in model.layers:
layer.trainable = state
model.trainable = state
class RandomWeightedAverage(_Merge):
"""Takes a randomly-weighted average of two tensors. In geometric terms, this outputs a random point on the line
between each pair of input points.
Inheriting from _Merge is a little messy but it was the quickest solution I could think of.
Improvements appreciated."""
def _merge_function(self, inputs):
weights = K.random_uniform((BATCH_SIZE, 1))
print(inputs[0])
return (weights * inputs[0]) + ((1 - weights) * inputs[1])
# The generator_model is used when we want to train the generator layers.
# As such, we ensure that the discriminator layers are not trainable.
# Note that once we compile this model, updating .trainable will have no effect within it. As such, it
# won't cause problems if we later set discriminator.trainable = True for the discriminator_model, as long
# as we compile the generator_model first.
set_trainable_state(discriminator, False)
set_trainable_state(generator, True)
# enable_train = Input(shape = (1,))
generator_input = Input(shape=generator_input_shape)
generator_layers = generator(generator_input)
# masked_generator_layers = Lambda(mul)(generator_layers)
# discriminator_noise_in = Input(shape=(1,))
# input_seq_g = Input(shape = discriminator_input_shape2)
discriminator_layers_for_generator= discriminator([generator_layers, generator_input])
generator_model = Model(inputs=[generator_input, enable_train],
outputs=[generator_layers, discriminator_layers_for_generator])
# We use the Adam paramaters from Gulrajani et al.
# global generator_recunstruction_loss_new
# generator_recunstruction_loss_new = partial(generator_recunstruction_loss, enableTrain = enable_train)
# generator_recunstruction_loss_new.__name__ = 'generator_RLN'
loss = [generator_recunstruction_loss_new, wasserstein_loss]
loss_weights = [0, 1]
generator_model.compile(optimizer=Adam(lr=1E-4, beta_1=0.9, beta_2=0.999, epsilon=1e-08),
loss=loss, loss_weights=loss_weights)
# Now that the generator_model is compiled, we can make the discriminator layers trainable.
set_trainable_state(discriminator, True)
set_trainable_state(generator, False)
# The discriminator_model is more complex. It takes both real image samples and random noise seeds as input.
# The noise seed is run through the generator model to get generated images. Both real and generated images
# are then run through the discriminator. Although we could concatenate the real and generated images into a
# single tensor, we don't (see model compilation for why).
real_samples = Input(shape=discriminator_input_shape)
input_seq = Input(shape=discriminator_input_shape2)
generator_input_for_discriminator = Input(shape=generator_input_shape)
generated_samples_for_discriminator = generator(generator_input_for_discriminator)
# masked_real_samples = Lambda(mul)(real_samples)
# masked_generated_samples_for_discriminator= Lambda(mul)(generated_samples_for_discriminator)
discriminator_output_from_generator = discriminator(
[generated_samples_for_discriminator, generator_input_for_discriminator])
discriminator_output_from_real_samples= discriminator([real_samples, input_seq])
# We also need to generate weighted-averages of real and generated samples, to use for the gradient norm penalty.
averaged_samples = RandomWeightedAverage()([real_samples, generated_samples_for_discriminator])
average_seq = RandomWeightedAverage()([input_seq, generator_input_for_discriminator])
# We then run these samples through the discriminator as well. Note that we never really use the discriminator
# output for these samples - we're only running them to get the gradient norm for the gradient penalty loss.
# print('hehehe')
# print(averaged_samples)
# masked_averaged_samples = Lambda(mul)(averaged_samples)
averaged_samples_out = discriminator([averaged_samples, average_seq])
# The gradient penalty loss function requires the input averaged samples to get gradients. However,
# Keras loss functions can only have two arguments, y_true and y_pred. We get around this by making a partial()
# of the function with the averaged samples here.
partial_gp_loss = partial(gradient_penalty_loss,
averaged_samples=averaged_samples,
gradient_penalty_weight=GRADIENT_PENALTY_WEIGHT)
partial_gp_loss.__name__ = 'gradient_penalty' # Functions need names or Keras will throw an error
# Keras requires that inputs and outputs have the same number of samples. This is why we didn't concatenate the
# real samples and generated samples before passing them to the discriminator: If we had, it would create an
# output with 2 * BATCH_SIZE samples, while the output of the "averaged" samples for gradient penalty
# would have only BATCH_SIZE samples.
# If we don't concatenate the real and generated samples, however, we get three outputs: One of the generated
# samples, one of the real samples, and one of the averaged samples, all of size BATCH_SIZE. This works neatly!
discriminator_model = Model(inputs=[real_samples, generator_input_for_discriminator, input_seq],
outputs=[discriminator_output_from_real_samples,
discriminator_output_from_generator,
averaged_samples_out])
loss_weights2 = [1, 1, 1]
# We use the Adam paramaters from Gulrajani et al. We use the Wasserstein loss for both the real and generated
# samples, and the gradient penalty loss for the averaged samples.
discriminator_model.compile(optimizer=SGD(clipvalue=0.01), # Adam(lr=4E-4, beta_1=0.9, beta_2=0.999, epsilon=1e-08)
loss=[wasserstein_loss,
wasserstein_loss,
partial_gp_loss], loss_weights=loss_weights2)
# set_trainable_state(discriminator, True)
# set_trainable_state(generator, True)
return generator_model, discriminator_model

+ 868
- 0
pfp_conditional_wgan_exp1.py View File

#!/usr/bin/env python

"""

"""
from __future__ import division

import logging
import sys
import time
from collections import deque
from multiprocessing import Pool

import click as ck
import numpy as np
import pandas as pd
import tensorflow as tf
from keras import backend as K
from keras.callbacks import EarlyStopping, ModelCheckpoint
from keras.layers import (
Dense, Input, SpatialDropout1D, Conv1D, MaxPooling1D, AveragePooling1D, Multiply,GaussianNoise,
Flatten, Concatenate, Add, Maximum, Embedding, BatchNormalization, Activation, Dropout)
from keras.losses import binary_crossentropy
from keras.models import Sequential, Model, load_model
from keras.preprocessing import sequence
from scipy.spatial import distance
from sklearn.metrics import log_loss
from sklearn.metrics import roc_curve, auc, matthews_corrcoef
from keras.layers import Lambda, LeakyReLU, RepeatVector
from sklearn.metrics import precision_recall_curve
from keras.callbacks import TensorBoard

from utils import (
get_gene_ontology,
get_go_set,
get_anchestors,
get_parents,
DataGenerator,
FUNC_DICT,
get_height)
from conditional_wgan_wrapper_exp1 import WGAN_wrapper, wasserstein_loss, generator_recunstruction_loss_new

config = tf.ConfigProto()
config.gpu_options.allow_growth = True
sess = tf.Session(config=config)
K.set_session(sess)

logging.basicConfig(format='%(levelname)s:%(message)s', level=logging.INFO)
sys.setrecursionlimit(100000)

DATA_ROOT = 'Data/data1/'
MAXLEN = 1000
REPLEN = 256
ind = 0


@ck.command()
@ck.option(
'--function',
default='bp',
help='Ontology id (mf, bp, cc)')
@ck.option(
'--device',
default='gpu:0',
help='GPU or CPU device id')
@ck.option(
'--org',
default= None,
help='Organism id for filtering test set')
@ck.option('--train', default=True, is_flag=True)

def main(function, device, org, train):
global FUNCTION
FUNCTION = function
global GO_ID
GO_ID = FUNC_DICT[FUNCTION]
global go
go = get_gene_ontology('go.obo')
global ORG
ORG = org
func_df = pd.read_pickle(DATA_ROOT + FUNCTION + '.pkl')
global functions
functions = func_df['functions'].values
global embedds
embedds = load_go_embedding(functions)
global all_functions
all_functions = get_go_set(go, GO_ID)
logging.info('Functions: %s %d' % (FUNCTION, len(functions)))
if ORG is not None:
logging.info('Organism %s' % ORG)
global node_names
node_names = set()
with tf.device('/' + device):
params = {
'fc_output': 1024,
'embedding_dims': 128,
'embedding_dropout': 0.2,
'nb_conv': 2,
'nb_dense': 1,
'filter_length': 128,
'nb_filter': 32,
'pool_length': 64,
'stride': 32
}
model(params, is_train=train)


def load_go_embedding(functions):
f_embedding = open('data/swiss/term-graph-cellular-component-100d.emb', 'r')
f_index = open('data/swiss/term-id-cellular-component.txt','r')
embeddings = f_embedding.readlines()
index = f_index.readlines()
dict1 ={}
for i in range(1, len(index)):
ind = map(str, index[i].split())
dict1[ind[0]] = ind[2]
dict2={}
for i in range(1, len(embeddings)):
embedding = map(str, embeddings[i].split())
dict2[str(embedding[0])] = embedding[1:]

embedds = []

z = 0
for i in range(len(functions)):
term = functions[i]
if term in dict1.keys():
term_index = dict1[term]
term_embedd = dict2[str(term_index)]
else:
term_embedd = np.ones([100])
embedds.append(term_embedd)
return np.array(embedds)

def load_data():

df = pd.read_pickle(DATA_ROOT + 'train' + '-' + FUNCTION + '.pkl')
n = len(df)
index = df.index.values
valid_n = int(n * 0.8)
train_df = df.loc[index[:valid_n]]
valid_df = df.loc[index[valid_n:]]
test_df = pd.read_pickle(DATA_ROOT + 'test' + '-' + FUNCTION + '.pkl')
print( test_df['orgs'] )
if ORG is not None:
logging.info('Unfiltered test size: %d' % len(test_df))
test_df = test_df[test_df['orgs'] == ORG]
logging.info('Filtered test size: %d' % len(test_df))

def reshape(values):
values = np.hstack(values).reshape(
len(values), len(values[0]))
return values

def normalize_minmax(values):
mn = np.min(values)
mx = np.max(values)
if mx - mn != 0.0:
return (values - mn) / (mx - mn)
return values - mn

def get_values(data_frame):
print(data_frame['labels'].values.shape)
labels = reshape(data_frame['labels'].values)
ngrams = sequence.pad_sequences(
data_frame['ngrams'].values, maxlen=MAXLEN)
ngrams = reshape(ngrams)
rep = reshape(data_frame['embeddings'].values)
data = ngrams
return data, labels

train = get_values(train_df)
valid = get_values(valid_df)
test = get_values(test_df)

return train, valid, test, train_df, valid_df, test_df


def load_data0():

df = pd.read_pickle(DATA_ROOT + 'train' + '-' + FUNCTION + '.pkl')
n = len(df)
index = df.index.values
valid_n = int(n * 0.8)
train_df = df.loc[index[:valid_n]]
valid_df = df.loc[index[valid_n:]]
test_df = pd.read_pickle(DATA_ROOT + 'test' + '-' + FUNCTION + '.pkl')
print( test_df['orgs'] )
if ORG is not None:
logging.info('Unfiltered test size: %d' % len(test_df))
test_df = test_df[test_df['orgs'] == ORG]
logging.info('Filtered test size: %d' % len(test_df))

def reshape(values):
values = np.hstack(values).reshape(
len(values), len(values[0]))
return values

def normalize_minmax(values):
mn = np.min(values)
mx = np.max(values)
if mx - mn != 0.0:
return (values - mn) / (mx - mn)
return values - mn

def get_values(data_frame, p_list):
accesion_numbers = data_frame['accessions'].values
all_labels = data_frame['labels'].values
all_ngrams = reshape(sequence.pad_sequences(data_frame['ngrams'].values, maxlen=MAXLEN) )
print(len(all_labels))
print(len(p_list))
labels = []
data = []
for i in range(accesion_numbers.shape[0]):
if accesion_numbers[i] in p_list:
labels.append(all_labels[i])
data.append(all_ngrams[i])

return np.asarray(data), np.asarray(labels)

dev_F = pd.read_csv('dev-' + FUNCTION + '-input.tsv', sep='\t', header=0, index_col=0)
test_F = pd.read_csv('test-' + FUNCTION + '-input.tsv', sep='\t', header=0, index_col=0)
train_F = pd.read_csv('train-' + FUNCTION + '-input.tsv', sep='\t', header=0, index_col=0)

train_l = train_F.index.tolist()
test_l = test_F.index.tolist()
dev_l = dev_F.index.tolist()

train = get_values(train_df, train_l)
valid = get_values(valid_df, dev_l)
test = get_values(test_df, test_l)

return train, valid, test, train_df, valid_df, test_df





class TrainValTensorBoard(TensorBoard):
def __init__(self, log_dir='./t_logs', **kwargs):
# Make the original `TensorBoard` log to a subdirectory 'training'
training_log_dir = log_dir
super(TrainValTensorBoard, self).__init__(training_log_dir, **kwargs)

# Log the validation metrics to a separate subdirectory
self.val_log_dir = './v_logs'
self.dis_log_dir = './d_logs'
self.gen_log_dir = './g_logs'

def set_model(self, model):
# Setup writer for validation metrics
self.val_writer = tf.summary.FileWriter(self.val_log_dir)
self.dis_writer = tf.summary.FileWriter(self.dis_log_dir)
self.gen_writer = tf.summary.FileWriter(self.gen_log_dir)
super(TrainValTensorBoard, self).set_model(model)

def on_epoch_end(self, epoch, logs=None):
# Pop the validation logs and handle them separately with
# `self.val_writer`. Also rename the keys so that they can
# be plotted on the same figure with the training metrics
logs = logs or {}
val_logs = {k.replace('val_', 'v_'): v for k, v in logs.items() if k.startswith('val_')}
for name, value in val_logs.items():
summary = tf.Summary()
summary_value = summary.value.add()
summary_value.simple_value = value.item()
summary_value.tag = name
self.val_writer.add_summary(summary, epoch)
self.val_writer.flush()


logs = logs or {}
dis_logs = {k.replace('discriminator_', 'd_'): v for k, v in logs.items() if k.startswith('discriminator_')}
for name, value in dis_logs.items():
summary = tf.Summary()
summary_value = summary.value.add()
summary_value.simple_value = value.item()
summary_value.tag = name
self.dis_writer.add_summary(summary, epoch)
self.dis_writer.flush()

logs = logs or {}
gen_logs = {k.replace('generator_', 'g_'): v for k, v in logs.items() if k.startswith('generator_')}
for name, value in gen_logs.items():
summary = tf.Summary()
summary_value = summary.value.add()
summary_value.simple_value = value.item()
summary_value.tag = name
self.gen_writer.add_summary(summary, epoch)
self.gen_writer.flush()

# Pass the remaining logs to `TensorBoard.on_epoch_end`
t_logs = {k: v for k, v in logs.items() if not k.startswith('val_')}
tr_logs = {k: v for k, v in t_logs.items() if not k.startswith('discriminator_')}
tra_logs = {k: v for k, v in tr_logs.items() if not k.startswith('generator_')}
super(TrainValTensorBoard, self).on_epoch_end(epoch, tra_logs)

def on_train_end(self, logs=None):
super(TrainValTensorBoard, self).on_train_end(logs)
self.val_writer.close()
self.gen_writer.close()
self.dis_writer.close()


def get_feature_model(params):
embedding_dims = params['embedding_dims']
max_features = 8001
model = Sequential()
model.add(Embedding(
max_features,
embedding_dims,
input_length=MAXLEN))
model.add(SpatialDropout1D(0.4))
model.add(Conv1D(
padding="valid",
strides=1,
filters=params['nb_filter'],
kernel_size=params['filter_length']))
model.add(LeakyReLU())
#model.add(BatchNormalization())
model.add(Conv1D(
padding="valid",
strides=1,
filters=params['nb_filter'],
kernel_size=params['filter_length']))
model.add(LeakyReLU())
# model.add(BatchNormalization()) problem in loading the best model
model.add(AveragePooling1D(strides=params['stride'], pool_size=params['pool_length']))
model.add(Flatten())
return model



#def get_node_name(go_id, unique=False):
# name = go_id.split(':')[1]
# if not unique:
# return name
# if name not in node_names:
# node_names.add(name)
# return name
# i = 1
# while (name + '_' + str(i)) in node_names:
# i += 1
# name = name + '_' + str(i)
# node_names.add(name)
# return name


#def get_layers(inputs):
# q = deque()
# layers = {}
# name = get_node_name(GO_ID)
# layers[GO_ID] = {'net': inputs}
# for node_id in go[GO_ID]['children']:
# if node_id in func_set:
# q.append((node_id, inputs))
# while len(q) > 0:
# node_id, net = q.popleft()
# name = get_node_name(node_id)
# net, output = get_function_node(name, inputs)
# if node_id not in layers:
# layers[node_id] = {'net': net, 'output': output}
# for n_id in go[node_id]['children']:
# if n_id in func_set and n_id not in layers:
# ok = True
# for p_id in get_parents(go, n_id):
# if p_id in func_set and p_id not in layers:
# ok = False
# if ok:
# q.append((n_id, net))

# for node_id in functions:
# childs = set(go[node_id]['children']).intersection(func_set)
# if len(childs) > 0:
# outputs = [layers[node_id]['output']]
# for ch_id in childs:
# outputs.append(layers[ch_id]['output'])
# name = get_node_name(node_id) + '_max'
# layers[node_id]['output'] = Maximum(name=name)(outputs)
# return layers

#def get_function_node(name, inputs):
# output_name = name + '_out'
# output = Dense(1, name=output_name, activation='sigmoid')(inputs)
# return output, output


def get_generator(params, n_classes):
inputs = Input(shape=(MAXLEN,), dtype='int32', name='input1')
feature_model = get_feature_model(params)(inputs)
net = Dense(200)(feature_model) #bp(dis1 g1) 400 - cc 200 - mf 200
net = BatchNormalization()(net)
net = LeakyReLU()(net)
#To impose DAG hierarchy directly to the model (inspired by DeepGO)
#layers = get_layers(net)
#output_models = []
#for i in range(len(functions)):
# output_models.append(layers[functions[i]]['output'])
#net = Concatenate(axis=1)(output_models)
output = Dense(n_classes, activation='tanh')(net) #'sigmoid'
model = Model(inputs=inputs, outputs=output)
return model


def get_discriminator(params, n_classes, dropout_rate=0.5):
inputs = Input(shape=(n_classes, ))
#n_inputs = GaussianNoise(0.1)(inputs)
#x = Conv1D(filters=8, kernel_size=128, padding='valid', strides=1)(n_inputs)
#x = BatchNormalization()(x)
#x = LeakyReLU()(x)
#x = Conv1D(filters=8, kernel_size=128, padding='valid', strides=1)(x)
#x = BatchNormalization()(x)
#x = LeakyReLU()(x)
#x = AveragePooling1D(strides=params['stride'], pool_size=params['pool_length'])(x)
#x = Flatten()(x)
#x = Lambda(lambda y: K.squeeze(y, 2))(x)
inputs2 = Input(shape =(MAXLEN,), dtype ='int32', name='d_input2')
x2 = Embedding(8001,128, input_length=MAXLEN)(inputs2)
x2 = SpatialDropout1D(0.4)(x2)
x2 = Conv1D(filters=8, kernel_size=128, padding='valid', strides=1)(x2)
x2 = BatchNormalization()(x2)
x2 = LeakyReLU()(x2)
#x2 = Conv1D(filters=1, kernel_size=1, padding='valid', strides=1)(x2)
#x2 = BatchNormalization()(x2)
#x2 = LeakyReLU()(x2)
#x2 = AveragePooling1D(strides=params['stride'], pool_size=params['pool_length'])(x2)
x2 = Flatten()(x2)

size = 100
x2 = Dropout(dropout_rate)(x2)
x2 = Dense(size)(x2)
x2 = BatchNormalization()(x2)
x2 = LeakyReLU()(x2)

size = 100
x = Dropout(dropout_rate)(inputs)
x = Dense(size)(x)
x = BatchNormalization()(x)
x = LeakyReLU()(x)

x = Concatenate(axis=1, name='merged2')([x, x2])
layer_sizes = [30, 30, 30, 30, 30, 30]
for size in layer_sizes:
x = Dropout(dropout_rate)(x)
x = Dense(size)(x)
x = BatchNormalization()(x)
x = LeakyReLU()(x)

outputs = Dense(1)(x)
model = Model(inputs = [inputs, inputs2], outputs=[outputs], name='Discriminator')
return model


def rescale_labels(labels):
rescaled_labels = 2*(labels - 0.5)
return rescaled_labels


def rescale_back_labels(labels):
rescaled_back_labels =(labels/2)+0.5
return rescaled_back_labels


def get_model(params,nb_classes, batch_size, GRADIENT_PENALTY_WEIGHT=10):
generator = get_generator(params, nb_classes)
discriminator = get_discriminator(params, nb_classes)

generator_model, discriminator_model = \
WGAN_wrapper(generator=generator,
discriminator=discriminator,
generator_input_shape=(MAXLEN,),
discriminator_input_shape=(nb_classes,),
discriminator_input_shape2 = (MAXLEN, ),
batch_size=batch_size,
gradient_penalty_weight=GRADIENT_PENALTY_WEIGHT,
embeddings=embedds)

logging.info('Compilation finished')
return generator_model, discriminator_model


def train_wgan(generator_model, discriminator_model, batch_size, epochs,
x_train, y_train, x_val, y_val, x_test, y_test, generator_model_path, discriminator_model_path,
TRAINING_RATIO=10, N_WARM_UP=20):
BATCH_SIZE = batch_size
N_EPOCH = epochs

positive_y = np.ones((batch_size, 1), dtype=np.float32)
zero_y = positive_y * 0
negative_y = -positive_y
positive_full_y = np.ones((BATCH_SIZE * TRAINING_RATIO, 1), dtype=np.float32)
dummy_y = np.zeros((BATCH_SIZE, 1), dtype=np.float32)

positive_full_enable_train = np.ones((len(x_train), 1), dtype=np.float32)
positive_full_enable_val = np.ones((len(x_val), 1), dtype=np.float32)
positive_full_enable_test = np.ones((len(x_test), 1), dtype=np.float32)
best_validation_loss = None

callback = TrainValTensorBoard(write_graph=False) #TensorBoard(log_path)
callback.set_model(generator_model)

for epoch in range(N_EPOCH):
# np.random.shuffle(X_train)
print("Epoch: ", epoch)
print("Number of batches: ", int(y_train.shape[0] // BATCH_SIZE))
discriminator_loss = []
generator_loss = []
minibatches_size = BATCH_SIZE * TRAINING_RATIO

shuffled_indexes = np.random.permutation(x_train.shape[0])
shuffled_indexes_2 = np.random.permutation(x_train.shape[0])

for i in range(int(y_train.shape[0] // (BATCH_SIZE * TRAINING_RATIO))):
batch_indexes = shuffled_indexes[i * minibatches_size:(i + 1) * minibatches_size]
batch_indexes_2 = shuffled_indexes_2[i * minibatches_size:(i + 1) * minibatches_size]
x = x_train[batch_indexes]
y = y_train[batch_indexes]
y_2 = y_train[batch_indexes_2]
x_2 = x_train[batch_indexes_2]
if epoch < N_WARM_UP:
for j in range(TRAINING_RATIO):
x_batch = x[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]
y_batch = y[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]

generator_loss.append(generator_model.train_on_batch([x_batch, positive_y], [y_batch, zero_y]))
discriminator_loss.append(0)
else:
for j in range(TRAINING_RATIO):
x_batch = x[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]
y_batch = y[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]
y_batch_2 = y_2[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]
x_batch_2 = x_2[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]
# noise = np.random.rand(BATCH_SIZE, 100).astype(np.float32)
noise = x_batch
discriminator_loss.append(discriminator_model.train_on_batch(
[y_batch_2, noise, x_batch_2],
[positive_y, negative_y, dummy_y]))

generator_loss.append(generator_model.train_on_batch([x, positive_full_y], [y, positive_full_y]))

predicted_y_train = generator_model.predict([x_train, positive_full_enable_train], batch_size=BATCH_SIZE)[0]
predicted_y_val = generator_model.predict([x_val, positive_full_enable_val], batch_size=BATCH_SIZE)[0]

train_loss = log_loss(rescale_back_labels(y_train), rescale_back_labels(predicted_y_train))
val_loss = log_loss(y_val, rescale_back_labels(predicted_y_val))
callback.on_epoch_end(epoch,
logs={'discriminator_loss': (np.sum(np.asarray(discriminator_loss)) if discriminator_loss else -1),
'generator_loss': (np.sum(np.asarray(generator_loss)) if generator_loss else -1),#train_loss,
'train_bce_loss': train_loss,
'val_bce_loss': val_loss})

print("train bce loss: {:.4f}, validation bce loss: {:.4f}, generator loss: {:.4f}, discriminator loss: {:.4f}".format(
train_loss, val_loss,
(np.sum(np.asarray(generator_loss)) if generator_loss else -1) / x_train.shape[0],
(np.sum(np.asarray(discriminator_loss)) if discriminator_loss else -1) / x_train.shape[0]))

if best_validation_loss is None or best_validation_loss > val_loss:
print('\nEpoch %05d: improved from %0.5f,'
' saving model to %s and %s'
% (epoch + 1, val_loss, generator_model_path, discriminator_model_path))

best_validation_loss = val_loss
generator_model.save(generator_model_path, overwrite=True)
discriminator_model.save(discriminator_model_path, overwrite=True)


def model(params, batch_size=64, nb_epoch=10000, is_train=True):
nb_classes = len(functions)

logging.info("Loading Data")
train, val, test, train_df, valid_df, test_df = load_data()
train_data, train_labels = train
val_data, val_labels = val
test_data, test_labels = test
train_labels = rescale_labels(train_labels)
logging.info("Training data size: %d" % len(train_data))
logging.info("Validation data size: %d" % len(val_data))
logging.info("Test data size: %d" % len(test_data))
generator_model_path = DATA_ROOT + 'models/new_model_seq_' + FUNCTION + '.h5'
discriminator_model_path = DATA_ROOT + 'models/new_model_disc_seq_' + FUNCTION + '.h5'

logging.info('Starting training the model')
train_generator = DataGenerator(batch_size, nb_classes)
train_generator.fit(train_data, train_labels)
valid_generator = DataGenerator(batch_size, nb_classes)
valid_generator.fit(val_data, val_labels)
test_generator = DataGenerator(batch_size, nb_classes)
test_generator.fit(test_data, test_labels)

if is_train:
generator_model, discriminator_model = get_model(params, nb_classes, batch_size)
#Loading a pretrained model
#generator_model.load_weights(DATA_ROOT + 'models/trained_bce/new_model_seq_' + FUNCTION + '.h5')

train_wgan(generator_model, discriminator_model, batch_size=batch_size, epochs=nb_epoch,
x_train=train_data, y_train=train_labels , x_val=val_data, y_val=val_labels, x_test = test_data, y_test = test_labels,
generator_model_path=generator_model_path,
discriminator_model_path=discriminator_model_path)

logging.info('Loading best model')
model = load_model(generator_model_path,
custom_objects={'generator_recunstruction_loss_new': generator_recunstruction_loss_new,
'wasserstein_loss': wasserstein_loss})

logging.info('Predicting')
start_time = time.time()
preds = model.predict_generator(test_generator, steps=len(test_data) / batch_size)[0]
end_time = time.time()
logging.info('%f prediction time for %f protein sequences' % (start_time-end_time, len(test_data)))

preds = rescale_back_labels(preds)
#preds = propagate_score(preds, functions)
logging.info('Computing performance')
f, p, r, t, preds_max = compute_performance(preds, test_labels)
micro_roc_auc = compute_roc(preds, test_labels)
macro_roc_auc = compute_roc_macro(preds, test_labels)
mcc = compute_mcc(preds_max, test_labels)
aupr, _ = compute_aupr(preds, test_labels)
tpr_score, total = compute_tpr_score(preds_max, functions)


m_pr_max, m_rc_max, m_f1_max, M_pr_max, M_rc_max, M_f1_max = micro_macro_function_centric_f1(preds.T, test_labels.T)
function_centric_performance(functions, preds.T, test_labels.T, rescale_back_labels(train_labels).T, t)

logging.info('Protein centric macro Th, PR, RC, F1: \t %f %f %f %f' % (t, p, r, f))
logging.info('Micro ROC AUC: \t %f ' % (micro_roc_auc, ))
logging.info('Macro ROC AUC: \t %f ' % (macro_roc_auc,))
logging.info('MCC: \t %f ' % (mcc, ))
logging.info('AUPR: \t %f ' % (aupr, ))
logging.info('TPR_Score from total: \t %f \t %f ' % (tpr_score,total))
logging.info('Function centric macro PR, RC, F1: \t %f %f %f' % (M_pr_max, M_rc_max, M_f1_max) )
logging.info('Function centric micro PR, RC, F1: \t %f %f %f' % (m_pr_max, m_rc_max, m_f1_max) )

def function_centric_performance(functions, preds, labels, labels_train, threshold):
results = []
preds = np.round(preds, 2)
for i in range(preds.shape[0]):
predictions = (preds[i, :] > threshold).astype(np.int32)
tp = np.sum(predictions * labels[i, :])
fp = np.sum(predictions) - tp
fn = np.sum(labels[i, :]) - tp
if tp > 0:
precision = tp / (1.0 * (tp + fp))
recall = tp / (1.0 * (tp + fn))
f = 2 * precision * recall / (precision + recall)
else:
if fp == 0 and fn == 0:
precision = 1
recall = 1
f = 1
else:
precision = 0
recall = 0
f = 0
f_max = f
p_max = precision
r_max = recall
num_prots_train = np.sum(labels_train[i, :])
height = get_height(go, functions[i])
results.append([functions[i], num_prots_train, height, f_max, p_max, r_max])
results = pd.DataFrame(results)
results.to_csv('Con_GodGanSeq_results_' + FUNCTION + '.txt', sep='\t', index=False)


def micro_macro_function_centric_f1(preds, labels):
predictions = np.round(preds)
m_tp = 0
m_fp = 0
m_fn = 0
M_pr = 0
M_rc = 0
for i in range(preds.shape[0]):
tp = np.sum(predictions[i, :] * labels[i, :])
fp = np.sum(predictions[i, :]) - tp
fn = np.sum(labels[i, :]) - tp
m_tp += tp
m_fp += fp
m_fn += fn
if tp > 0:
pr = 1.0 * tp / (1.0 * (tp + fp))
rc = 1.0 * tp / (1.0 * (tp + fn))
else:
if fp == 0 and fn == 0:
pr = 1
rc = 1
else:
pr = 0
rc = 0
M_pr += pr
M_rc += rc

if m_tp > 0:
m_pr = 1.0 * m_tp / (1.0 * (m_tp + m_fp))
m_rc = 1.0 * m_tp / (1.0 * (m_tp + m_fn))
m_f1 = 2.0 * m_pr * m_rc / (m_pr + m_rc)
else:
if m_fp == 0 and m_fn == 0:
m_pr = 1
m_rc = 1
m_f1 = 1
else:
m_pr = 0
m_rc = 0
m_f1 = 0

M_pr /= preds.shape[0]
M_rc /= preds.shape[0]
if M_pr == 0 and M_rc == 0:
M_f1 = 0
else:
M_f1 = 2.0 * M_pr * M_rc / (M_pr + M_rc)

return m_pr, m_rc, m_f1, M_pr, M_rc, M_f1


def compute_roc(preds, labels):
# Compute ROC curve and ROC area for each class
fpr, tpr, _ = roc_curve(labels.flatten(), preds.flatten())
roc_auc = auc(fpr, tpr)
return roc_auc


def compute_roc_macro(yhat_raw, y):
if yhat_raw.shape[0] <= 1:
return
fpr = {}
tpr = {}
roc_auc = {}
# get AUC for each label individually
relevant_labels = []
auc_labels = {}
for i in range(y.shape[1]):
# only if there are true positives for this label
if y[:, i].sum() > 0:
fpr[i], tpr[i], _ = roc_curve(y[:, i], yhat_raw[:, i])
if len(fpr[i]) > 1 and len(tpr[i]) > 1:
auc_score = auc(fpr[i], tpr[i])
if not np.isnan(auc_score):
auc_labels["auc_%d" % i] = auc_score
relevant_labels.append(i)

# macro-AUC: just average the auc scores
aucs = []
for i in relevant_labels:
aucs.append(auc_labels['auc_%d' % i])
roc_auc_macro = np.mean(aucs)
return roc_auc_macro



def compute_aupr(preds, labels):
# Compute ROC curve and ROC area for each class
pr, rc, threshold =precision_recall_curve(labels.flatten(), preds.flatten())
pr_auc = auc(rc, pr)
M_pr_auc = 0
return pr_auc, M_pr_auc

def compute_mcc(preds, labels):
# Compute ROC curve and ROC area for each class
mcc = matthews_corrcoef(labels.flatten(), preds.flatten())
return mcc


def compute_performance(preds, labels):
predicts = np.round(preds, 2)
f_max = 0
p_max = 0
r_max = 0
t_max = 0
for t in range(1, 100):
threshold = t / 100.0
predictions = (predicts > threshold).astype(np.int32)
total = 0
f = 0.0
p = 0.0
r = 0.0
p_total = 0
for i in range(labels.shape[0]):
tp = np.sum(predictions[i, :] * labels[i, :])
fp = np.sum(predictions[i, :]) - tp
fn = np.sum(labels[i, :]) - tp
if tp == 0 and fp == 0 and fn == 0:
continue
total += 1
if tp != 0:
p_total += 1
precision = tp / (1.0 * (tp + fp))
recall = tp / (1.0 * (tp + fn))
p += precision
r += recall
if p_total == 0:
continue
r /= total
p /= p_total
if p + r > 0:
f = 2 * p * r / (p + r)
if f_max < f:
f_max = f
p_max = p
r_max = r
t_max = threshold
predictions_max = predictions
return f_max, p_max, r_max, t_max, predictions_max



def get_anchestors(go, go_id):
go_set = set()
q = deque()
q.append(go_id)
while(len(q) > 0):
g_id = q.popleft()
go_set.add(g_id)
for parent_id in go[g_id]['is_a']:
if parent_id in go:
q.append(parent_id)
return go_set

def compute_tpr_score(preds_max, functions):
my_dict = {}
for i in range(len(functions)):
my_dict[functions[i]] = i
total =0
anc = np.zeros([len(functions), len(functions)])
for i in range(len(functions)):
anchestors = list(get_anchestors(go, functions[i]))
for j in range(len(anchestors)):
if anchestors[j] in my_dict.keys():
anc[i, my_dict[anchestors[j]]]=1
total +=1



tpr_score = 0
for i in range(preds_max.shape[0]):
for j in range(preds_max.shape[1]):
if preds_max[i, j] == 1:
for k in range(anc.shape[1]):
if anc[j, k]==1:
if preds_max[i,k]!=1:
tpr_score = tpr_score +1

return tpr_score/preds_max.shape[0], total


def propagate_score(preds, functions):
my_dict = {}
for i in range(len(functions)):
my_dict[functions[i]] = i

anc = np.zeros([len(functions), len(functions)])
for i in range(len(functions)):
anchestors = list(get_anchestors(go, functions[i]))
for j in range(len(anchestors)):
if anchestors[j] in my_dict.keys():
anc[i, my_dict[anchestors[j]]] = 1

new_preds = np.array(preds)
for i in range(new_preds.shape[1]):
for k in range(anc.shape[1]):
if anc[i, k] == 1:
new_preds[new_preds[:, i] > new_preds[:, k], k] = new_preds[new_preds[:, i] > new_preds[:, k], i]

return new_preds


if __name__ == '__main__':
main()

+ 636
- 0
pfp_conditional_wgan_exp2.py View File

#!/usr/bin/env python

"""

"""
from __future__ import division

import logging
import sys
import time
import click as ck
import numpy as np
import pandas as pd
import tensorflow as tf
from keras import backend as K
from keras.layers import (
Dense, Input, SpatialDropout1D, Conv1D, MaxPooling1D,GaussianNoise,
Flatten, Concatenate, Add, Maximum, Embedding, BatchNormalization, Activation, Dropout)
from keras.models import Sequential, Model, load_model
from sklearn.metrics import log_loss
from sklearn.metrics import roc_curve, auc, matthews_corrcoef, mean_squared_error
from keras.layers import Lambda, LeakyReLU
from sklearn.metrics import precision_recall_curve
from keras.callbacks import TensorBoard
from skimage import measure

from utils import DataGenerator
from conditional_wgan_wrapper_exp2 import WGAN_wrapper, wasserstein_loss, generator_recunstruction_loss_new

config = tf.ConfigProto()
config.gpu_options.allow_growth = True
sess = tf.Session(config=config)
K.set_session(sess)

logging.basicConfig(format='%(levelname)s:%(message)s', level=logging.INFO)
sys.setrecursionlimit(100000)

DATA_ROOT = 'Data/data2/'
MAXLEN = 258


@ck.command()
@ck.option(
'--function',
default='cc',
help='Ontology id (mf, bp, cc)')
@ck.option(
'--device',
default='gpu:0',
help='GPU or CPU device id')
@ck.option('--train',default =True, is_flag=True)

def main(function, device, train):
global FUNCTION
FUNCTION = function
with tf.device('/' + device):
params = {
'fc_output': 1024,
'embedding_dims': 128,
'embedding_dropout': 0.2,
'nb_conv': 1,
'nb_dense': 1,
'filter_length': 128,
'nb_filter': 32,
'pool_length': 64,
'stride': 32
}
model(params, is_train=train)



def load_data2():
all_data_x_fn = 'data2/all_data_X.csv'
all_data_x = pd.read_csv(all_data_x_fn, sep='\t', header=0, index_col=0)
all_proteins_train = [p.replace('"', '') for p in all_data_x.index]
all_data_x.index = all_proteins_train
all_data_y_fn = 'data2/all_data_Y.csv'
all_data_y = pd.read_csv(all_data_y_fn, sep='\t', header=0, index_col=0)
branch = pd.read_csv('data2/'+FUNCTION +'_branches.txt', sep='\t', header=0, index_col=0)
all_x = all_data_x.values
branches = [p for p in branch.index.tolist() if p in all_data_y.columns.tolist()]
branches = list(dict.fromkeys(branches))
global mt_funcs
mt_funcs = branches
t= pd.DataFrame(all_data_y, columns=branches)
all_y = t.values
number_of_test = int(np.ceil(0.4 * len(all_x)))
np.random.seed(0)
index = np.random.rand(1,number_of_test)
index_test = [int(p) for p in np.ceil(index*(len(all_x)-1))[0]]
index_train = [p for p in range(len(all_x)) if p not in index_test]
train_data = all_x[index_train, : ]
test_x = all_x[index_test, : ]
train_labels = all_y[index_train, : ]
test_y = all_y[index_test, :]

number_of_vals = int(np.ceil(0.5 * len(test_x)))
np.random.seed(1)
index = np.random.rand(1, number_of_vals)
index_vals = [int(p) for p in np.ceil(index * (len(test_x) - 1))[0]]
index_test = [p for p in range(len(test_x)) if p not in index_vals]
val_data = test_x[index_vals, :]
test_data = test_x[index_test, :]
val_labels = test_y[index_vals, :]
test_labels = test_y[index_test, :]

print(train_labels.shape)
return train_data, train_labels, test_data, test_labels, val_data, val_labels


def get_feature_model(params):
embedding_dims = params['embedding_dims']
max_features = 8001
model = Sequential()
model.add(Embedding(
max_features,
embedding_dims,
input_length=MAXLEN))
model.add(SpatialDropout1D(0.4))
for i in range(params['nb_conv']):
model.add(Conv1D(
padding="valid",
strides=1,
filters=params['nb_filter'],
kernel_size=params['filter_length']))
model.add(LeakyReLU())
model.add(MaxPooling1D(strides=params['stride'], pool_size=params['pool_length']))
model.add(Flatten())
return model


def get_generator(params, n_classes):
inputs = Input(shape=(MAXLEN,), dtype='float32', name='input1')
net0 = Dense(200)(inputs)
#net0 = BatchNormalization()(net0)
net0 = LeakyReLU()(net0)
net0 = Dense(200)(net0)
#net0 = BatchNormalization()(net0)
net0 = LeakyReLU()(net0)
#net0 = Dense(200, activation='relu')(inputs)
net0 = Dense(200)(net0)
#net0 = BatchNormalization()(net0)
net = LeakyReLU()(net0)
output = Dense(n_classes, activation='tanh')(net)
model = Model(inputs=inputs, outputs=output)
return model


def get_discriminator(params, n_classes, dropout_rate=0.5):
inputs = Input(shape=(n_classes, ))
n_inputs = GaussianNoise(0.1)(inputs)
inputs2 = Input(shape =(MAXLEN,), dtype ='int32', name='d_input2')
x2 = Embedding(8001, 128, input_length=MAXLEN)(inputs2)
x2 = Conv1D(filters =1 , kernel_size= 1, padding = 'valid', strides=1)(x2)
x2 = LeakyReLU()(x2)
x2 = Lambda(lambda x: K.squeeze(x, 2))(x2)
#for i in range(params['nb_conv']):
# x2 = Conv1D ( activation="relu", padding="valid", strides=1, filters=params['nb_filter'],kernel_size=params['filter_length'])(x2)


#x2 =MaxPooling1D(strides=params['stride'], pool_size=params['pool_length'])(x2)
#x2 = Flatten()(x2)
size = 40
x = n_inputs
x = Dropout(dropout_rate)(x)
x = Dense(size)(x)
x = BatchNormalization()(x)
x = LeakyReLU()(x)

size = 40
x2 = Dropout(dropout_rate)(x2)
x2 = Dense(size)(x2)
x2 = BatchNormalization()(x2)
x2 = LeakyReLU()(x2)

x = Concatenate(axis =1 , name = 'merged2')([x, x2])
layer_sizes = [30, 30, 30, 30, 30, 30]
for size in layer_sizes:
x = Dropout(dropout_rate)(x)
x = Dense(size)(x)
x = BatchNormalization()(x)
x = LeakyReLU()(x)

outputs = Dense(1)(x)
model = Model(inputs = [inputs ,inputs2], outputs=outputs, name='Discriminator')
return model

class TrainValTensorBoard(TensorBoard):
def __init__(self, log_dir='./t_logs', **kwargs):
# Make the original `TensorBoard` log to a subdirectory 'training'
training_log_dir = log_dir
super(TrainValTensorBoard, self).__init__(training_log_dir, **kwargs)

# Log the validation metrics to a separate subdirectory
self.val_log_dir = './v_logs'
self.dis_log_dir = './d_logs'
self.gen_log_dir = './g_logs'

def set_model(self, model):
# Setup writer for validation metrics
self.val_writer = tf.summary.FileWriter(self.val_log_dir)
self.dis_writer = tf.summary.FileWriter(self.dis_log_dir)
self.gen_writer = tf.summary.FileWriter(self.gen_log_dir)
super(TrainValTensorBoard, self).set_model(model)

def on_epoch_end(self, epoch, logs=None):
# Pop the validation logs and handle them separately with
# `self.val_writer`. Also rename the keys so that they can
# be plotted on the same figure with the training metrics
logs = logs or {}
val_logs = {k.replace('val_', 'v_'): v for k, v in logs.items() if k.startswith('val_')}
for name, value in val_logs.items():
summary = tf.Summary()
summary_value = summary.value.add()
summary_value.simple_value = value.item()
summary_value.tag = name
self.val_writer.add_summary(summary, epoch)
self.val_writer.flush()

logs = logs or {}
dis_logs = {k.replace('discriminator_', 'd_'): v for k, v in logs.items() if k.startswith('discriminator_')}
for name, value in dis_logs.items():
summary = tf.Summary()
summary_value = summary.value.add()
summary_value.simple_value = value.item()
summary_value.tag = name
self.dis_writer.add_summary(summary, epoch)
self.dis_writer.flush()

logs = logs or {}
gen_logs = {k.replace('generator_', 'g_'): v for k, v in logs.items() if k.startswith('generator_')}
for name, value in gen_logs.items():
summary = tf.Summary()
summary_value = summary.value.add()
summary_value.simple_value = value.item()
summary_value.tag = name
self.gen_writer.add_summary(summary, epoch)
self.gen_writer.flush()

# Pass the remaining logs to `TensorBoard.on_epoch_end`
t_logs = {k: v for k, v in logs.items() if not k.startswith('val_')}
tr_logs = {k: v for k, v in t_logs.items() if not k.startswith('discriminator_')}
tra_logs = {k: v for k, v in tr_logs.items() if not k.startswith('generator_')}
super(TrainValTensorBoard, self).on_epoch_end(epoch, tra_logs)

def on_train_end(self, logs=None):
super(TrainValTensorBoard, self).on_train_end(logs)
self.val_writer.close()
self.gen_writer.close()
self.dis_writer.close()


def rescale_labels(labels):
rescaled_labels = 2*(labels - 0.5)
return rescaled_labels


def rescale_back_labels(labels):
rescaled_back_labels =(labels/2)+0.5
return rescaled_back_labels


def get_model(params, nb_classes, batch_size, GRADIENT_PENALTY_WEIGHT=10):
generator = get_generator(params, nb_classes)
discriminator = get_discriminator(params, nb_classes)

generator_model, discriminator_model = \
WGAN_wrapper(generator=generator,
discriminator=discriminator,
generator_input_shape=(MAXLEN,),
discriminator_input_shape=(nb_classes,),
discriminator_input_shape2 = (MAXLEN, ),
batch_size=batch_size,
gradient_penalty_weight=GRADIENT_PENALTY_WEIGHT)

logging.info('Compilation finished')
return generator_model, discriminator_model


def train_wgan(generator_model, discriminator_model, batch_size, epochs,
x_train, y_train, x_val, y_val, generator_model_path, discriminator_model_path,
TRAINING_RATIO=10, N_WARM_UP=0):
BATCH_SIZE = batch_size
N_EPOCH = epochs

positive_y = np.ones((batch_size, 1), dtype=np.float32)
zero_y = positive_y * 0
negative_y = -positive_y
positive_full_y = np.ones((BATCH_SIZE * TRAINING_RATIO, 1), dtype=np.float32)
dummy_y = np.zeros((BATCH_SIZE, 1), dtype=np.float32)
positive_full_enable_train = np.ones((len(x_train), 1), dtype = np.float32 )
positive_full_enable_val = np.ones((len(x_val), 1), dtype =np.float32 )
#positive_enable_train = np.ones((1, batch_size),dtype = np.float32 )
#positive_full_train_enable = np.ones((1,BATCH_SIZE * TRAINING_RATIO ), dtype=np.float32 )
best_validation_loss = None
callback = TrainValTensorBoard(write_graph=False) #TensorBoard(log_path)
callback.set_model(generator_model)

for epoch in range(N_EPOCH):
# np.random.shuffle(X_train)
print("Epoch: ", epoch)
print("Number of batches: ", int(y_train.shape[0] // BATCH_SIZE))
discriminator_loss = []
generator_loss = []
minibatches_size = BATCH_SIZE * TRAINING_RATIO

shuffled_indexes = np.random.permutation(x_train.shape[0])
shuffled_indexes_2 = np.random.permutation(x_train.shape[0])

for i in range(int(y_train.shape[0] // (BATCH_SIZE * TRAINING_RATIO))):
batch_indexes = shuffled_indexes[i * minibatches_size:(i + 1) * minibatches_size]
batch_indexes_2 = shuffled_indexes_2[i * minibatches_size:(i + 1) * minibatches_size]
x = x_train[batch_indexes]
y = y_train[batch_indexes]
y_2 = y_train[batch_indexes_2]
x_2 = x_train[batch_indexes_2]
if epoch < N_WARM_UP:
for j in range(TRAINING_RATIO):
x_batch = x[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]
y_batch = y[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]

logg = generator_model.train_on_batch([x_batch, positive_y], [y_batch, zero_y])
generator_loss.append(logg)
discriminator_loss.append(0)
else:
for j in range(TRAINING_RATIO):
x_batch = x[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]
y_batch_2 = y_2[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]
x_batch_2 = x_2[j * BATCH_SIZE:(j + 1) * BATCH_SIZE]
# noise = np.random.rand(BATCH_SIZE, 100).astype(np.float32)
noise = x_batch
discriminator_loss.append(discriminator_model.train_on_batch(
[y_batch_2, noise, x_batch_2],
[positive_y, negative_y, dummy_y]))

logg = generator_model.train_on_batch([x,positive_full_y], [y, positive_full_y])
generator_loss.append(logg)
predicted_y_train, _ = generator_model.predict([x_train, positive_full_enable_train], batch_size=BATCH_SIZE)
predicted_y_val, _ = generator_model.predict([x_val, positive_full_enable_val], batch_size=BATCH_SIZE)

train_loss = log_loss(rescale_back_labels(y_train), rescale_back_labels(predicted_y_train))
val_loss = log_loss(y_val, rescale_back_labels(predicted_y_val))
callback.on_epoch_end(epoch, logs={'discriminator_loss': (np.sum(np.asarray(discriminator_loss)) if discriminator_loss else -1),
'generator_loss': (np.sum(np.asarray(generator_loss)) if generator_loss else -1) ,'train_loss':train_loss ,'val_loss': val_loss})
print("train loss: {:.4f}, validation loss: {:.4f}, validation auc: {:.4f}, generator loss: {:.4f}, discriminator loss: {:.4f}".format(
train_loss, val_loss, compute_roc(rescale_back_labels(predicted_y_val), y_val),
(np.sum(np.asarray(generator_loss)) if generator_loss else -1) / x_train.shape[0],
(np.sum(np.asarray(discriminator_loss)) if discriminator_loss else -1) / x_train.shape[0]))

if best_validation_loss is None or best_validation_loss > val_loss:
print('\nEpoch %05d: improved from %0.5f,'
' saving model to %s and %s'
% (epoch + 1, val_loss, generator_model_path, discriminator_model_path))

best_validation_loss = val_loss
generator_model.save(generator_model_path, overwrite=True)
discriminator_model.save(discriminator_model_path, overwrite=True)


def model(params, batch_size=128, nb_epoch=50, is_train=True):
start_time = time.time()

logging.info("Loading Data")
train_data, train_labels, test_data, test_labels, val_data, val_labels = load_data2()
train_labels = rescale_labels(train_labels)
nb_classes = train_labels.shape[1]
logging.info("Data loaded in %d sec" % (time.time() - start_time))
logging.info("Training data size: %d" % len(train_data))
logging.info("Validation data size: %d" % len(val_data))
logging.info("Test data size: %d" % len(test_data))
generator_model_path = DATA_ROOT + 'models/new_model_seq_mt_10000_1_loaded_pretrained_new_' + FUNCTION + '.h5' #_loaded_pretrained_new
discriminator_model_path = DATA_ROOT + 'models/new_model_disc_seq_mt_' + FUNCTION + '.h5'

logging.info('Starting training the model')
train_generator = DataGenerator(batch_size, nb_classes)
train_generator.fit(train_data, train_labels)
valid_generator = DataGenerator(batch_size, nb_classes)
valid_generator.fit(val_data, val_labels)
test_generator = DataGenerator(batch_size, nb_classes)
test_generator.fit(test_data, test_labels)

if is_train:
generator_model, discriminator_model = get_model(params, nb_classes, batch_size)

generator_model.load_weights(DATA_ROOT + 'models/new_model_seq_mt_1_0_' + FUNCTION + '.h5')

train_wgan(generator_model, discriminator_model, batch_size=batch_size, epochs=nb_epoch,
x_train=train_data, y_train=train_labels, x_val=val_data, y_val=val_labels,
generator_model_path=generator_model_path,
discriminator_model_path=discriminator_model_path)

logging.info('Loading best model')
model = load_model(generator_model_path,
custom_objects={'generator_recunstruction_loss_new': generator_recunstruction_loss_new,
'wasserstein_loss': wasserstein_loss})

logging.info('Predicting')
preds = model.predict_generator(test_generator, steps=len(test_data) / batch_size)[0]
preds = rescale_back_labels(preds)

logging.info('Computing performance')
f, p, r, t, preds_max = compute_performance(preds, test_labels)
roc_auc = compute_roc(preds, test_labels)
mcc = compute_mcc(preds_max, test_labels)
aupr , _ = compute_aupr(preds, test_labels)
m_pr_max, m_rc_max, m_f1_max, M_pr_max, M_rc_max, M_f1_max = micro_macro_function_centric_f1(preds.T, test_labels.T)
tpr, total_rule = tpr_score(preds_max)

logging.info('Protein centric macro Th, PR, RC, F1: \t %f %f %f %f' % (t, p, r, f))
logging.info('ROC AUC: \t %f ' % (roc_auc, ))
logging.info('MCC: \t %f ' % (mcc, ))
logging.info('AUPR: \t %f ' % (aupr, ))
logging.info('TPR from total rule: \t %f \t %f' % (tpr, total_rule))
logging.info('Function centric macro PR, RC, F1: \t %f %f %f' % (M_pr_max, M_rc_max, M_f1_max) )
logging.info('Function centric micro PR, RC, F1: \t %f %f %f' % (m_pr_max, m_rc_max, m_f1_max) )

function_centric_performance(preds.T, test_labels.T, rescale_back_labels(train_labels).T, t)


def tpr_score(preds):
hierarchy_f = open("data2/mtdnn_data_svmlight/" + FUNCTION + "_hierarchy.txt", "r")
hiers = hierarchy_f.readlines()
hiers.pop(0)
tree = []
for i in range(len(hiers)):
sample = []
str_sample = hiers[i].split()
for j in range(1, len(str_sample)):
sample.append(int(str_sample[j]))
tree.append(sample)
my_dict = {}
for i in range(preds.shape[0]):
my_dict[str(i)] = []
total_rule = 0
for i in range(len(tree)):
for j in range(len(tree[i])):
my_dict[str(tree[i][j]-1)].append(i)
total_rule +=1

tpr = 0
for i in range(preds.shape[0]):
for j in range(preds.shape[1]):
if preds[i,j] ==1:
parents = my_dict[str(j)]
for k in range(len(parents)):
if preds[i, parents[k]] == 0:
tpr = tpr +1

return tpr/preds.shape[0], total_rule

def function_centric_performance(preds, labels, labels_train, threshold):
results = []
av_f = 0
preds = np.round(preds, 2)
for i in range(preds.shape[0]):
predictions = (preds[i, :] > threshold).astype(np.int32)
tp = np.sum(predictions * labels[i, :])
fp = np.sum(predictions) - tp
fn = np.sum(labels[i, :]) - tp
if tp > 0:
precision = tp / (1.0 * (tp + fp))
recall = tp / (1.0 * (tp + fn))
f = 2 * precision * recall / (precision + recall)
else:
if fp == 0 and fn == 0:
precision = 1
recall = 1
f = 1
else:
precision = 0
recall = 0
f = 0
f_max = f
av_f += f
p_max = precision
r_max = recall
num_prots_train = np.sum(labels_train[i, :])
results.append([mt_funcs[i], num_prots_train, f_max, p_max, r_max])
print(av_f/preds.shape[0])
results = pd.DataFrame(results)
results.to_csv('MT_1_0_results' + FUNCTION + '.txt', sep='\t', index=False)

predictions = (preds > threshold).astype(np.int32)
hitmap_test = (np.dot(predictions, predictions.T)> 0).astype(np.int32)
hitmap_train =(np.dot(labels_train, labels_train.T) > 0).astype(np.int32)

logging.info('Heatmap scores:\t MSE:%f \t SSIM %f ' % (mean_squared_error(hitmap_test, hitmap_train), measure.compare_ssim(hitmap_test, hitmap_train)))

#hitmap_test = pd.DataFrame(hitmap_test)
#hitmap_test.to_csv('hitmap_1_1_' + FUNCTION + '.txt', sep='\t', index=False)

#hitmap_train = pd.DataFrame(hitmap_train)
#hitmap_train.to_csv('hitmap_train_' + FUNCTION + '.txt', sep='\t', index=False)

#co_occure = np.zeros(shape=[predictions.shape[0], predictions.shape[0]])
#mutual_ex = np.zeros(shape=[predictions.shape[0], predictions.shape[0]])

#check_co_occur_test = np.zeros(shape=[predictions.shape[0], predictions.shape[0]])

#for i in range(predictions.shape[0]):
# for j in range(predictions.shape[0]):
# temp = np.logical_xor(labels_train[i,:], labels_train[j,:])
# if np.sum(temp) ==0:
# co_occure[i, j] = 1
# temp = np.dot(labels_train[i,:], labels_train[j,:])
# if np.sum(temp) ==0:
# mutual_ex[i,j] = 1
# check_co_occur_test[i, j] = (sum(np.logical_xor(labels_train[i,:], labels_train[j,:])) >0).astype(np.int32)

#logging.info('Mutual_exclucivity_distorsion:\t %f ' % (sum(sum(np.multiply(mutual_ex, hitmap_test)))))
#logging.info('Co_occurance_distorsion:\t %f ' % (sum(sum(np.multiply(co_occure, check_co_occur_test)))))


def micro_macro_function_centric_f1(preds, labels):
predictions = np.round(preds)
m_tp = 0
m_fp = 0
m_fn = 0
M_pr = 0
M_rc = 0
for i in range(preds.shape[0]):
tp = np.sum(predictions[i, :] * labels[i, :])
fp = np.sum(predictions[i, :]) - tp
fn = np.sum(labels[i, :]) - tp
m_tp += tp
m_fp += fp
m_fn += fn
if tp > 0:
pr = 1.0 * tp / (1.0 * (tp + fp))
rc = 1.0 * tp / (1.0 * (tp + fn))
else:
if fp == 0 and fn == 0:
pr = 1
rc = 1
else:
pr = 0
rc = 0
M_pr += pr
M_rc += rc

if m_tp > 0:
m_pr = 1.0 * m_tp / (1.0 * (m_tp + m_fp))
m_rc = 1.0 * m_tp / (1.0 * (m_tp + m_fn))
m_f1 = 2.0 * m_pr * m_rc / (m_pr + m_rc)
else:
if m_fp == 0 and m_fn == 0:
m_pr = 1
m_rc = 1
m_f1 = 1
else:
m_pr = 0
m_rc = 0
m_f1 = 0

M_pr /= preds.shape[0]
M_rc /= preds.shape[0]
if M_pr == 0 and M_rc == 0:
M_f1 = 0
else:
M_f1 = 2.0 * M_pr * M_rc / (M_pr + M_rc)

return m_pr, m_rc, m_f1, M_pr, M_rc, M_f1

def compute_roc(preds, labels):
# Compute ROC curve and ROC area for each class
fpr, tpr, _ = roc_curve(labels.flatten(), preds.flatten())
roc_auc = auc(fpr, tpr)
return roc_auc


def compute_aupr(preds, labels):
# Compute ROC curve and ROC area for each class
pr, rc, threshold =precision_recall_curve(labels.flatten(), preds.flatten())
pr_auc = auc(rc, pr)
#pr, rc, threshold =precision_recall_curve(labels.flatten(), preds.flatten(),average ='macro' )
M_pr_auc = 0
return pr_auc, M_pr_auc


def compute_mcc(preds, labels):
# Compute ROC curve and ROC area for each class
mcc = matthews_corrcoef(labels.flatten(), preds.flatten())
return mcc

def compute_performance(preds, labels):
preds = np.round(preds, 2)
f_max = 0
p_max = 0
r_max = 0
t_max = 0
for t in range(1, 100):
threshold = t / 100.0
predictions = (preds > threshold).astype(np.int32)
total = 0
f = 0.0
p = 0.0
r = 0.0
p_total = 0
for i in range(labels.shape[0]):
tp = np.sum(predictions[i, :] * labels[i, :])
fp = np.sum(predictions[i, :]) - tp
fn = np.sum(labels[i, :]) - tp
if tp == 0 and fp == 0 and fn == 0:
continue
total += 1
if tp != 0:
p_total += 1
precision = tp / (1.0 * (tp + fp))
recall = tp / (1.0 * (tp + fn))
p += precision
r += recall
if p_total == 0:
continue
r /= total
p /= p_total
if p + r > 0:
f = 2 * p * r / (p + r)
if f_max < f:
f_max = f
p_max = p
r_max = r
t_max = threshold
predictions_max = predictions
return f_max, p_max, r_max, t_max, predictions_max


if __name__ == '__main__':
main()

+ 5
- 0
utils.py View File

# These functions are used to manipulate first dataset from DeepGO and taken from https://github.com/bio-ontology-research-group/deepgo.


from collections import deque from collections import deque
from keras import backend as K from keras import backend as K
from keras.callbacks import ModelCheckpoint from keras.callbacks import ModelCheckpoint
import pandas as pd import pandas as pd
from xml.etree import ElementTree as ET from xml.etree import ElementTree as ET




BIOLOGICAL_PROCESS = 'GO:0008150' BIOLOGICAL_PROCESS = 'GO:0008150'
MOLECULAR_FUNCTION = 'GO:0003674' MOLECULAR_FUNCTION = 'GO:0003674'
CELLULAR_COMPONENT = 'GO:0005575' CELLULAR_COMPONENT = 'GO:0005575'

Write Preview
Loading…
Cancel
Save