search

GRAPH-0 / JODO

Learning Joint 2D & 3D Diffusion Models for Complete Molecule Generation
MIT License
40 stars 8 forks source link

Other dataset apply #4

Closed yddr closed 10 months ago

yddr commented 10 months ago

Hi, Thanks for releasing awesome code!

I want to apply to other dataset not QM9, GEOM, and so on. I converted smiles dataset to .sdf file. But I'm confused the our processed dataset files you mentioned in readme.md file.

Well, your preprocess method is datasets/build_data.py from what I understand, right? And can I use the other dataset, if use the build_dataset.py?

I'll really appreciate for your reply.

GRAPH-0 commented 10 months ago

Unfortunately, we do not currently offer support for applying to other datasets. But I may help you implement your dataset within our framework. This mainly involves creating a new XX_dataset.py and modifying the related contents of build_dataset.py and datasets_config.py. If you need to generate 2D graphs, please refer to the zinc_dataset.py file. On the other hand, if you need to generate 3D geometric graphs, you can refer to the qm9_dataset.py file for implementation. In fact, the preprocessing files we provide are mainly implemented by the process() function in InMemoryDataset, except for Geom.

yddr commented 10 months ago

I'll try implement with referencing your reply and files.

Thanks for the quick and kind reply.