Guolin Ke

Thank you very much! I receive the data. it seems the data don't include template part (`template_mmcif_dir` and `mmcif_cache.json`), are they not needed?

This is a good question. As described in our preprint (Sec 2.2), we make several modifications to the model architecture, and thus, our model cannot convert back to the original...

I think it can't be either. Uni-Fold-Jax uses the same model architecture as AlphaFold.

Hi @sokrypton I saw the CUDA version is 11.2 > | NVIDIA-SMI 460.32.03 Driver Version: 460.32.03 CUDA Version: 11.2 while pytorch is `torch.__version__ = 1.12.0+cu113`, which is cuda 11.3. Therefore,...

we just add a colab version of Uni-Fold, https://colab.research.google.com/github/dptech-corp/Uni-Fold/blob/main/notebooks/unifold.ipynb

The dataset is very large, and we are looking for a solution for data hosting. Last week we submitted the request to "AWS Open Data Sponsorship Application", but didn't receive...

The multimer features mostly are the same as monomer ones, except the assembly of multiple chains. You can refer this script https://github.com/dptech-corp/Uni-Fold/blob/main/scripts/get_pdb_assembly.py to generate the "pdb_assembly.json" we used.

@shiyu1994 can you help to reivew?

@shiyu1994 do you have a windows machine to test? I think our CI should covert the msvc build, and it shouldn't fail.

will this affect the training speed? any benchmarks?