Question about rescoring (Fig. 5 in the paper)

[nonsensejoke]

Hi there! 👋

I am writing to you regarding your excellent article on EquiScore, "Generic protein–ligand interaction scoring by integrating physical prior knowledge and data augmentation modelling," published in Nature Machine Intelligence. I was particularly impressed by the model’s robust rescoring capabilities shown in Figure 5.

My question relates to the specific implementation of rescoring. EquiScore is a binary classification model that outputs a probability (prob) assessing the binding potential. When EquiScore was used to enhance the virtual screening performance of various docking methods, could you clarify whether or not that you have explored generating a hybrid score by multiplying the EquiScore prediction probability by the original docking score?

Thank you for your time and effort in addressing this technical question. I appreciate your valuable contribution to the field of structure-based drug design.

[CAODH]

in our paper, we only used the scores from our model to rescoring ,but i thank using a hybrid score is a very valuable idea> Hi there! 👋

I am writing to you regarding your excellent article on EquiScore, "Generic protein–ligand interaction scoring by integrating physical prior knowledge and data augmentation modelling," published in Nature Machine Intelligence. I was particularly impressed by the model’s robust rescoring capabilities shown in Figure 5.

My question relates to the specific implementation of rescoring. EquiScore is a binary classification model that outputs a probability (prob) assessing the binding potential. When EquiScore was used to enhance the virtual screening performance of various docking methods, could you clarify whether or not that you have explored generating a hybrid score by multiplying the EquiScore prediction probability by the original docking score?

Thank you for your time and effort in addressing this technical question. I appreciate your valuable contribution to the field of structure-based drug design.