Why the screening Not accurate?

[SanFran-Me]

I generate the .pm file for rec.pdb and its ligand ref_lig.sdf
Then I put the ref_lig.sdf in the library, and run the screening code.
I expect that it will be in the top rank, but it comes out the ref_lig.sdf is in the 70,000th of 1500,000 (top 6.7%), which is not accurate enough as I wanted.
How could I fine tune the model if needed?

[SeonghwanSeo]

We use simple analytic function for high generalization, so screening performance may not be superior for all targets.

The model training was done in unsupervised way on public PDBBind2020 dataset, so fine-tuning may not be available (if you have your own private data, you can do it) even though I release the training code. (Sorry for the delayed release of the training script)

Primarily, you can tune the weight for each non-covalent interaction type for your target.

python screening.py --anion <ANION> --cation <CATION> --aromatic <AROMATIC> --hbd <HBD> --hba <HBA> --halogen <HALOGEN> --hydrophobic <HYDROPHOBIC>