# ds_file, label_index, rst_file, use_emb, hidden_dim
ds_file = '../../../yeast/preprocessed/Supp-AB.tsv'
label_index = 2
rst_file = 'results/15k_onehot_cnn.txt'
sid1_index = 0
sid2_index = 1
if len(sys.argv) > 1:
ds_file, label_index, rst_file, use_emb, hidden_dim, n_epochs = sys.argv[
1:]
label_index = int(label_index)
use_emb = int(use_emb)
hidden_dim = int(hidden_dim)
n_epochs = int(n_epochs)
seq2t = s2t(emb_files[use_emb])
max_data = -1
limit_data = max_data > 0
raw_data = []
skip_head = True
x = None
count = 0
for line in tqdm(open(ds_file)):
if skip_head:
skip_head = False
continue
line = line.rstrip('\n').rstrip('\r').split('\t')
if id2index.get(line[sid1_index]) is None or id2index.get(
line[sid2_index]) is None:
n_epochs=50
# ds_file, label_index, rst_file, use_emb, hiddem_dim
ds_file = '../../../sun/preprocessed/Supp-AB.tsv'
label_index = 2
rst_file = 'results/15k_onehot_cnn.txt'
sid1_index = 0
sid2_index = 1
if len(sys.argv) > 1:
ds_file, label_index, rst_file, use_emb, hidden_dim, n_epochs = sys.argv[1:]
label_index = int(label_index)
use_emb = int(use_emb)
hidden_dim = int(hidden_dim)
n_epochs = int(n_epochs)
seq2t = s2t(emb_files[use_emb]) ####s2t???????????????
max_data = -1
limit_data = max_data > 0
raw_data = []
skip_head = True
x = None
count = 0
for line in tqdm(open(ds_file)):
if skip_head:
skip_head = False
continue
line = line.rstrip('\n').rstrip('\r').split('\t')
# 与原始蛋白质库比对，若蛋白质名称无法匹配，则删除
if id2index.get(line[sid1_index]) is None or id2index.get(line[sid2_index]) is None:
continue
from keras.optimizers import Adam, RMSprop
from sklearn.metrics import accuracy_score, roc_auc_score, precision_score, recall_score, average_precision_score
import sys
from tqdm import tqdm
from numpy import linalg as LA
import scipy
import numpy as np
import os
import tensorflow as tf
import keras.backend.tensorflow_backend as KTF
from utils import *
from models import *
seq_size = 1000
MAXLEN = 1000
seq2t = s2t('vec5_CTC.txt')
hidden_dim = 50
dim = seq2t.dim
epochs = 5
num_gpus = 1
batch_size = 200*num_gpus
steps = 1000
thres = '0'
option = 'seq'
tid = sys.argv[2]
embedding_file = sys.argv[1]
print("option: ", option, "threshold: ", thres)
model_file = f'model_rcnn_{tid}.h5'
n_epochs = 50
'''
# ds_file, label_index, rst_file, use_emb, hidden_dim
ds_file = 'Supp-AB.tsv'
label_index = 2
rst_file = 'results/15k_onehot_cnn.txt'
sid1_index = 0
sid2_index = 1
if len(sys.argv) > 1:
ds_file, label_index, rst_file, use_emb, hidden_dim, n_epochs = sys.argv[1:]
label_index = int(label_index)
use_emb = int(use_emb)
hidden_dim = int(hidden_dim)
n_epochs = int(n_epochs)
'''
seq2t = s2t(emb_files[3])
max_data = -1
limit_data = max_data > 0
raw_data = []
skip_head = True
x = None
count = 0
'''
for line in tqdm(open(ds_file)):
if skip_head:
skip_head = False
continue
line = line.rstrip('\n').rstrip('\r').split('\t')
if id2index.get(line[sid1_index]) is None or id2index.get(line[sid2_index]) is None:
continue
