Novel interaction FP for CADD #chemoinfo

I am not familiar with CADD such as Docking, MD and ab initio calculation but interested in.
Recently I read novel interaction FP between protein and ligand reported in Journal of chemoinformatics. It shows good performance.
URL is below.
https://jcheminf.springeropen.com/track/pdf/10.1186/s13321-018-0264-0

The author propose novel FP protein per atom score contributions derived interaction fingerprint (PADIF) based on GOLD docking score.
It is interesting for me because this score based on protein per atom score contributions. It is indicated that user can analyze the score by atom based. So by using the score Chemist can design molecules focused on specific atoms (positions). In the report there are no data about that.

BYW the algorithm shows good performance against nuclear hormone receptors such as ER, GR and PPAR. I do not think that good docking score correlates ligand affinity but it is useful for pose prediction. And PADIF shows good performance for pose prediction.

How about use the score with MD calculation ??

Pose prediction is very important task for computer aid drug design but very difficult to predict correct pose. Because protein is too flexible, PDB reflects only snap shot of molecular dynamics.
I am looking forward to improvement of these area.