Inconsistent Environment Configuration and Dependency Version Compatibility

[HuaJiefang]

Hi！

Thank you for maintaining this project! I encountered some issues while setting up the environment and would appreciate your guidance:

Configuration Inconsistency
The dependencies listed in environment.yml do not fully align with the libraries mentioned in the README. This mismatch causes missing functionalities or errors after installation.
Version Compatibility Issues
When running conda env create -f environment.yml, version conflicts arise between certain libraries (e.g., LibraryA=1.0 and LibraryB=2.0 are incompatible), leading to installation failures.

Could you kindly confirm the following?

The exact CUDA version and Python version used for this project.
Whether you could share an updated/validated environment.yml that matches the README instructions.

This would greatly help the community reproduce and use your work smoothly. If additional details are needed (e.g., specific error logs), please let me know!

Thank you for your time and effort! 🙌

[Oldheaaa]

I encountered the same issue. Some packages require Python 3.10.0 or higher, but the current Python version is 3.9.19. This includes, but is not limited to, posecheck and sklearn.

[Brunxi]

Hello ! I am having the same issue here. Has anyone made any progress with this?

Thanks!

[Huajiefang258]

Hello ! I am having the same issue here. Has anyone made any progress with this?

Thanks!

yes,I installed packages, python and CUDA as the README narratives without utilization of environment.yml, but when using trained_model, an error occured:Segmentation fault, seems that the codes was packed falsely.

[Gao-RX412]

Hi, thanks for bringing this up!

The environment configuration has been updated in the README. Please refer to this for the updated setup instructions. Let us know if you run into any other issues.

[HuaJiefang]

Hi, thanks for bringing this up!嗨，谢谢你提出这个问题！

The environment configuration has been updated in the README. Please refer to this for the updated setup instructions. Let us know if you run into any other issues.README中的环境配置已经更新。请参考更新后的安装说明。如果你遇到任何其他问题，请告诉我们。

🙏 Thank you for addressing my previous issue!

I have a follow-up question regarding model inference:

Previously, I installed the library versions specified in the README. However, when attempting to generate molecules directly using the pre-trained model (without running the training step, only executing the command:
python -m scripts.sample_diffusion_no_pocket ../config/sampling/SMG/sample_diff_pos0_10_pos1.e-7_0.01_6_v001_bondTrue_scalar128_vec32_layer8_with_guidance.yml --data_id --result_path ./result/with_guide/),
I encountered an error： Segmentation fault.

💡 Could you please clarify:

Whether it is possible to use the trained model for molecule generation without running the training step.
How to specify a custom input molecule for generation (instead of using the default data index).
🛠️ If you could also check whether there might be issues in the project's packaging configuration (e.g., incomplete dependency installation or potential compatibility conflicts when using pre-built modules).

📋 Additionally, if you could update the README with more detailed instructions on inference workflows and configuration options, it would be greatly appreciated!

✨ Thank you for your time and support!

[ziqi92]

Hi,

Thank you for your interest in our work.

For the question,
(1) yes, it is definitely possible to use the trained model for molecule generation. Please follow the command to install the packages. If you cannot figure out which package has the compatibility issue, you may locate the issues with pdb or print, and let us know which function reports the error. Unfortunately, given the different machines, we cannot figure out which package has the compatibility issue in your case without more knowledge and information.

(2) to customize the test data, you can specify your test data in the config file below. Please make sure that the format follows the provided test data.

DiffSMol/config/sampling/SMG/sample_diff_pos0_10_pos1.e-7_0.01_6_v001_bondTrue_scalar128_vec32_layer8_no_guidance.yml

Line 9 in 5c276c0

path: ../data/MOSES2/MOSES2_test_mol.pkl

Thanks,
Ziqi