epochs = []
true_data_s = pd.DataFrame()
predict_data_s = pd.DataFrame()
for fold in range(n_kfold):
epoch, true_data, predict_data = mofgcn_new_target(
gene=gene,
cna=cna,
mutation=mutation,
drug_feature=feature_drug,
response_mat=cell_drug,
null_mask=null_mask,
target_dim=dim,
target_index=target_index,
evaluate_fun=roc_auc,
device="cuda:0")
true_data_s = true_data_s.append(translate_result(true_data))
predict_data_s = predict_data_s.append(
translate_result(predict_data))
epochs.append(epoch)
if dim:
file_drug.write(str(target_index) + ":" + str(epochs) + "\n")
true_data_s.to_csv("./result_data/drug_" + str(target_index) +
"_true_data.csv")
predict_data_s.to_csv("./result_data/drug_" + str(target_index) +
"_predict_data.csv")
else:
file_cell.write(str(target_index) + ":" + str(epochs) + "\n")
true_data_s.to_csv("./result_data/cell_" + str(target_index) +
"_true_data.csv")
predict_data_s.to_csv("./result_data/cell_" + str(target_index) +
"_predict_data.csv")
n_hid2=64,
alpha=8.70,
device="cuda:0")
opt = Optimizer(model,
sampler.train_data,
sampler.test_data,
sampler.test_mask,
sampler.train_mask,
roc_auc,
lr=1e-3,
epochs=1000,
device="cuda:0").to("cuda:0")
epoch, true_data, predict_data = opt()
epochs.append(epoch)
true_datas = true_datas.append(translate_result(true_data))
predict_datas = predict_datas.append(translate_result(predict_data))
file = open("./result_data/epochs.txt", "w")
file.write(str(epochs))
file.close()
pd.DataFrame(true_datas).to_csv("./result_data/true_data.csv")
pd.DataFrame(predict_datas).to_csv("./result_data/predict_data.csv")
"""
# 网格计算超参数，除alpha外
save_format = "{:^5d}{:^5d}{:^5d}{:^7d}{:^7d}{:7.2f}{:^9.5f}{:^9.4f}"
file = open("grid_result.txt", "w")
sigmas = [2, 3, 5, 7, 9]
knns = [2, 3, 5, 7, 9, 11]
iterates = [2, 3, 5, 7, 9]
n_hid1s = [36, 64, 128, 192]