环境问题

[QJYJH]

为什么我使用了你提供的surfgen的yml环境，运行“python gen.py --outdir example --check_point ./ckpt/val_119.pt --ply_file ./example/3cl_pocket_8.0_res_1.5.ply”这个命令时报错：Traceback (most recent call last):
File "gen.py", line 12, in
from utils.transforms import *
File "/home/data1/XRJ/SurfGen-main/utils/transforms.py", line 14, in
from torch_geometric.nn.pool import knn_graph
File "/home/data1/anaconda3/envs/geobind/lib/python3.7/site-packages/torch_geometric/init.py", line 4, in
import torch_geometric.data
File "/home/data1/anaconda3/envs/geobind/lib/python3.7/site-packages/torch_geometric/data/init.py", line 1, in
from .data import Data
File "/home/data1/anaconda3/envs/geobind/lib/python3.7/site-packages/torch_geometric/data/data.py", line 9, in
from torch_sparse import SparseTensor
ModuleNotFoundError: No module named 'torch_sparse'

[OdinZhang]

Dear Sir, the error suggests that you lack the torch_sparse, which may imply that your PyG suite is not successfully installed. My environment is built upon the CUDA 11.3, please manually config your environment according to your own CUDA version.

[QJYJH]

尊敬的先生，该错误表明您缺少torch_sparse，这可能意味着您的 PyG 套件未成功安装。我的环境是基于 CUDA 11.3 构建的，请根据您自己的 CUDA 版本手动配置您的环境。

Thank you so much for your reply, it solved the problem for me.I managed to generate some ligands of "3cl".I noticed that you used an "in situ estimation of binding energy—scoring energy" method in your validation experiments, can you specify how this was accomplished?

[OdinZhang]

Dear Sir, I have summarized the scripts at the https://github.com/HaotianZhangAI4Science/AI-Physics-DrugDiscovery/blob/main/pydocking/docking/qvina.py, the function scoring_with_qvina02 is exactly what you want.