[New Model] ProREM

[tyang816]

Hi, Pascal

We recently fused MSA into the ProSST model and achieved better performance, achieving a score of 0.518 on substitutions named ProREM. (https://arxiv.org/abs/2410.21127)

1. Added the scoring code of prorem in proteingym/baselines/prorem.
2. Added prorem information to config.json and constants.json.
3. Added scripts/scoring_DMS_zero_shot/scoring_ProREM_substitutions.sh.
4. Update the structure tokenizer of ProSST.

It is worth noting:

1. A new conda environment is required, the configuration file is in proteingym/baselines/prorem/environment.yaml (same as ProtSSN and ProSST), or you can use the environment of ProSST.
2. The original PDB files are same with ProtSSN and ProSST, you can download them here. The .a2m files can be downloaded here.
3. More model information can be seen in ProREM repository.

Thanks again for your community efforts.

best wishes
Yang Tan