You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I followed the default data preprocessing procedure in the original code and stored the pdbbind dataset in the form of a heterogeneous graph. However, when I read the heterogeneous graph file heterographs10.pkl, the following error occurred (as far as I know, this should correspond to the 1000 complexes data from 10000 to 11000 in the training set):
Traceback (most recent call last):
File "/data3/wangjs/workspace/DiffDock/train.py", line 227, in <module>
main_function()
File "/data3/wangjs/workspace/DiffDock/train.py", line 181, in main_function
train_loader, val_loader, val_dataset2 = construct_loader(args, t_to_sigma, device)
File "/data3/wangjs/workspace/DiffDock/datasets/loader.py", line 83, in construct_loader
train_dataset = PDBBind(cache_path=args.cache_path, split_path=args.split_train, keep_original=True,
File "/data3/wangjs/workspace/DiffDock/datasets/pdbbind.py", line 189, in __init__
self.complex_graphs, self.rdkit_ligands = self.collect_all_complexes()
File "/data3/wangjs/workspace/DiffDock/datasets/pdbbind.py", line 366, in collect_all_complexes
l = pickle.load(f)
RuntimeError: invalid value in pickle
It seems that the pkl file is completely corrupted, but I haven't been able to identify which complex or which part of the code caused this corruption.
Other heterogeneous graph files, such as heterographs9.pkl from pdbbind dataset, did not have this problem. After deleting heterographs10.pkl, I successfully reproduced the experimental results in your paper. I would like to know what happened and if there is a solution to this issue.
The text was updated successfully, but these errors were encountered:
I found that this RuntimeError occurred when trying to load rdkit_ligand10.pkl and that ligand in 3q4c was the cause. After excluding the preprocessing of 3q4c, I was able to run the DiffDock training
-----原始邮件-----
发件人:"Eric Chen" ***@***.***>
发送时间:2024-11-09 02:06:52 (星期六)
收件人: gcorso/DiffDock ***@***.***>
抄送: "A.硕" ***@***.***>, Author ***@***.***>
主题: Re: [gcorso/DiffDock] RuntimeError: invalid value in pickle (Issue #253)
I found that this RuntimeError occurred when trying to load rdkit_ligand10.pkl and that ligand in 3q4c was the cause. After excluding the preprocessing of 3q4c, I was able to run the DiffDock training
—
Reply to this email directly, view it on GitHub, or unsubscribe.
You are receiving this because you authored the thread.Message ID: ***@***.***>
I followed the default data preprocessing procedure in the original code and stored the pdbbind dataset in the form of a heterogeneous graph. However, when I read the heterogeneous graph file heterographs10.pkl, the following error occurred (as far as I know, this should correspond to the 1000 complexes data from 10000 to 11000 in the training set):
It seems that the pkl file is completely corrupted, but I haven't been able to identify which complex or which part of the code caused this corruption.
Other heterogeneous graph files, such as heterographs9.pkl from pdbbind dataset, did not have this problem. After deleting heterographs10.pkl, I successfully reproduced the experimental results in your paper. I would like to know what happened and if there is a solution to this issue.
The text was updated successfully, but these errors were encountered: