def evaluate():
with tf.Graph().as_default():
# 训练日志文件夹
logs_train_dir = '/'
# 验证文件夹
verification_dir = "/"
# 数据总数
n_test = 0
train, train_label = gbd.get_Data(verification_dir, ROW, COLUMN)
train_X, train_Y, one_hot_train_Y = gbd.get_batch_data(
train, train_label, batch_size=BATCH_SIZE)
train_logits, train_v_length = model.interface(
inputs=train_X,
Y=one_hot_train_Y,
batch_size=BATCH_SIZE,
vec_len=VEC_LEN,
temp_batch_size=BATCH_SIZE)
top_k_op = tf.nn.in_top_k(train_logits, train_Y, 1)
saver = tf.train.Saver(tf.global_variables())
with tf.Session() as sess:
print("Reading checkpoints...")
ckpt = tf.train.get_checkpoint_state(logs_train_dir)
if ckpt and ckpt.model_checkpoint_path:
global_step = ckpt.model_checkpoint_path.split('/')[-1].split(
'-')[-1]
saver.restore(sess, ckpt.model_checkpoint_path)
print("Loading success,global_step is %s" % global_step)
else:
print("no checkpoint file found")
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
try:
num_iter = int(n_test / BATCH_SIZE)
true_count = 0
total_sample_count = num_iter * BATCH_SIZE
step = 0
while step < num_iter and not coord.should_stop():
prediction = sess.run([top_k_op])
true_count += np.sum(prediction)
step += 1
precision = float(true_count) / total_sample_count
print("precision = %3f" % precision)
except Exception as e:
coord.request_stop(e)
finally:
coord.request_stop()
coord.join(threads)
def evaluate_one_base(ct):
"""
返回一条碱基的各部位对应(.)的概率
input:
ct: 文件夹，内部应只有一个ct文件，多了后果自负
outputs:
prediction: 二维张量，第一维因为batch_size = 1，一般索引数为碱基数；第二维索引数3，对应该部位为（ . ）的概率
"""
with tf.Graph().as_default():
train, train_label = gbd.get_Data(ct, ROW, COLUMN)
train_X, train_Y, one_hot_train_Y = gbd.get_batch_data(train, train_label, batch_size = BATCH_SIZE)
train_logits, train_v_length = model.interface(inputs = train_X,
Y = one_hot_train_Y,
batch_size = BATCH_SIZE,
vec_len = VEC_LEN,
temp_batch_size = BATCH_SIZE)
softmax = tf.nn.softmax(train_logits,dim=-1,name=None)
saver = tf.train.Saver(tf.global_variables())
with tf.Session() as sess:
print("Reading checkpoints...")
ckpt = tf.train.get_checkpoint_state(logs_train_dir)
if ckpt and ckpt.model_checkpoint_path:
global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1]
saver.restore(sess,ckpt.model_checkpoint_path)
print("Loading success,global_step is %s" % global_step)
else:
print("no checkpoint file found")
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess,coord = coord)
try:
if not coord.should_stop():
prediction = sess.run(softmax)
return prediction
except Exception as e:
coord.request_stop(e)
finally:
coord.request_stop()
coord.join(threads)
def run_trainning():
"""
对神经网络进行训练
"""
PATH = "Cleaned_5sRNA_test/"
logs_train_dir="Net_model/" # 保存训练得来的模型的文件夹
row = 19
column = 128
vec_len = 8
temp_batch_size = 64
inputs, Labels= gbd.get_Data(PATH=PATH,
row=row,
column=column)
train_X, train_Y, one_hot_train_Y = gbd.get_batch_data(inputs,Labels, batch_size=temp_batch_size)
train_logits,train_v_length=model.interface(inputs=train_X,
Y=one_hot_train_Y,
batch_size=temp_batch_size,
vec_len=vec_len,
temp_batch_size=temp_batch_size)
train_loss=model.loss(logits=train_logits,
v_length=train_v_length,
labels=train_Y,
Y=one_hot_train_Y,
temp_batch_size=temp_batch_size)
train_op = model.trainning(train_loss,learning_rate)
train_acc = model.evalution(train_logits,train_Y)
sess = tf.Session()
sess.run(tf.global_variables_initializer())
sess.run(tf.local_variables_initializer())
saver = tf.train.Saver()
coord = tf.train.Coordinator()
threads = tf.train.start_queue_runners(sess=sess,coord = coord)
try:
for step in np.arange(MAX_STEP):
if coord.should_stop():
break
_, tra_loss, tra_acc=sess.run([train_op, train_loss, train_acc])
if step % 100 == 0:
print("Step %d,train loss = %.2f,train accuracy = %.2f" %(step, tra_loss, tra_acc))
print(train_X.shape)
if step % 200 == 0:
# 每两百轮保存一次训练数据
checkpoint_path = os.path.join(logs_train_dir, 'model.ckpt')
saver.save(sess,
save_path=checkpoint_path,
global_step=step)
except tf.errors.OutOfRangeError:
print('Done Trainning')
finally:
coord.request_stop()
coord.join(threads)
sess.close()