def reset_basic_data(interaction_list, negative_interaction_list, lncRNA_list, protein_list):
for lncRNA in lncRNA_list:
lncRNA.interaction_list = []
for protein in protein_list:
protein.interaction_list = []
lncRNA_serial_number_index_dict = nodeSerialNumber_listIndex_dict_generation(lncRNA_list)
protein_serial_number_index_dict = nodeSerialNumber_listIndex_dict_generation(protein_list)
new_interaction_list = []
for i in range(len(interaction_list)):
interaction = interaction_list[i]
lncRNA_index = lncRNA_serial_number_index_dict[interaction.lncRNA.serial_number]
protein_index = protein_serial_number_index_dict[interaction.protein.serial_number]
temp_interaction = LncRNA_Protein_Interaction(lncRNA=lncRNA_list[lncRNA_index], protein=protein_list[protein_index], y=1)
new_interaction_list.append(temp_interaction)
lncRNA_list[lncRNA_index].interaction_list.append(temp_interaction)
protein_list[protein_index].interaction_list.append(temp_interaction)
new_negative_interaction_list = []
for i in range(len(negative_interaction_list)):
interaction = negative_interaction_list[i]
lncRNA_index = lncRNA_serial_number_index_dict[interaction.lncRNA.serial_number]
protein_index = protein_serial_number_index_dict[interaction.protein.serial_number]
temp_interaction = LncRNA_Protein_Interaction(lncRNA=lncRNA_list[lncRNA_index], protein=protein_list[protein_index], y=0)
new_negative_interaction_list.append(temp_interaction)
lncRNA_list[lncRNA_index].interaction_list.append(temp_interaction)
protein_list[protein_index].interaction_list.append(temp_interaction)
return new_interaction_list, new_negative_interaction_list, lncRNA_list, protein_list
def negative_interaction_generation():
global lncRNA_list, protein_list, interaction_list, negative_interaction_list, set_interactionKey, set_negativeInteractionKey
set_negativeInteractionKey = set()
if len(negative_interaction_list) != 0:
raise Exception('negative interactions exist')
num_of_interaction = len(interaction_list)
num_of_lncRNA = len(lncRNA_list)
num_of_protein = len(protein_list)
negative_interaction_count = 0
while (negative_interaction_count < num_of_interaction):
random_index_lncRNA = random.randint(0, num_of_lncRNA - 1)
random_index_protein = random.randint(0, num_of_protein - 1)
temp_lncRNA = lncRNA_list[random_index_lncRNA]
temp_protein = protein_list[random_index_protein]
# 检查随机选出的lncRNA和protein是不是有已知相互作用
key_negativeInteraction = (temp_lncRNA.serial_number,
temp_protein.serial_number)
if key_negativeInteraction in set_interactionKey:
continue
if key_negativeInteraction in set_negativeInteractionKey:
continue
# 经过检查，随机选出的lncRNA和protein是可以作为负样本的
set_negativeInteractionKey.add(key_negativeInteraction)
temp_interaction = LncRNA_Protein_Interaction(temp_lncRNA,
temp_protein, 0,
key_negativeInteraction)
negative_interaction_list.append(temp_interaction)
temp_lncRNA.interaction_list.append(temp_interaction)
temp_protein.interaction_list.append(temp_interaction)
negative_interaction_count = negative_interaction_count + 1
print('generate ', len(negative_interaction_list), ' negative samples')
def read_interaction_dataset(dataset_path, dataset_name):
global interaction_list, negative_interaction_list, lncRNA_list, protein_list, lncRNA_name_index_dict, protein_name_index_dict
# lncRNA_name_index_dict, protein_name_index_dic为了在interaction dataset中，读到重复的lncRNA或protein时
# 能在lncRNA_list和protein_list中快速的找到
if not osp.exists(dataset_path):
raise Exception('interaction datset does not exist')
print('start reading xlsx file')
wb = load_workbook(dataset_path)
sheets = wb.worksheets # 获取当前所有的sheet
sheet = sheets[0]
rows = sheet.rows
serial_number = 0
lncRNA_count = 0
protein_count = 0
flag = 0 #用来排除RPI7317.xlsx的第一行
for row in rows:
#排除第一行
if flag == 0:
flag = flag + 1
continue
#读出这一行的每个元素,每一行对应一个interaction实例，如果这个interaction对应的lncRNA和protein还没创建，就创建它
#并在索引词典中加入它在lncRNA_list或者protein_list中的索引
[lncRNA_name, protein_name, label] = [col.value for col in row]
label = int(label)
if lncRNA_name not in lncRNA_name_index_dict: # 新的，没创建过的lncRNA
temp_lncRNA = LncRNA(lncRNA_name, serial_number, 'LncRNA')
lncRNA_list.append(temp_lncRNA)
lncRNA_name_index_dict[lncRNA_name] = lncRNA_count
serial_number = serial_number + 1
lncRNA_count = lncRNA_count + 1
else: # 在interaction dataset中已经读到过，已经创建了对象的lncRNA，就存在lncRNA_list中
temp_lncRNA = lncRNA_list[lncRNA_name_index_dict[lncRNA_name]]
if protein_name not in protein_name_index_dict: # 新的，没创建过的protein
temp_protein = Protein(protein_name, serial_number, 'Protein')
protein_list.append(temp_protein)
protein_name_index_dict[protein_name] = protein_count
serial_number = serial_number + 1
protein_count = protein_count + 1
else: # 在interaction dataset中已经读到过，已经创建了对象的protein，就存在protein_list中
temp_protein = protein_list[protein_name_index_dict[protein_name]]
temp_interaction = LncRNA_Protein_Interaction(temp_lncRNA, temp_protein, label)
# print(temp_interaction.protein.name, temp_interaction.lncRNA.name)
if label == 1:
interaction_list.append(temp_interaction)
elif label == 0:
negative_interaction_list.append(temp_interaction)
else:
print(label)
raise Exception('{dataset_name}中有除了0和1之外的label'.format(dataset_name=dataset_name))
temp_lncRNA.interaction_list.append(temp_interaction)
temp_protein.interaction_list.append(temp_interaction)
print('number of lncRNA：{:d}, number of protein：{:d}, number of node：{:d}'.format(lncRNA_count, protein_count, lncRNA_count + protein_count))
print(f'number of positive samples：{len(interaction_list)}, number of negative samples：{len(negative_interaction_list)}\n')
def rebuild_all_negativeInteraction(set_negativeInteractionKey):
global lncRNA_list, protein_list, negative_interaction_list
dict_serialNumber_lncRNA = build_dict_serialNumber_node(lncRNA_list)
dict_serialNumber_protein = build_dict_serialNumber_node(protein_list)
# 根据set_negativeInteractionKey把负样本集构造出来
for negativeInteractionKey in set_negativeInteractionKey:
lncRNA_temp = dict_serialNumber_lncRNA[negativeInteractionKey[0]]
protein_temp = dict_serialNumber_protein[negativeInteractionKey[1]]
# 构造负样本
temp_negativeInteraction = LncRNA_Protein_Interaction(lncRNA_temp, protein_temp, 0, negativeInteractionKey)
negative_interaction_list.append(temp_negativeInteraction)
lncRNA_temp.interaction_list.append(temp_negativeInteraction)
protein_temp.interaction_list.append(temp_negativeInteraction)