I am writing to ask your opinion about the design of the ligand-binding site with Rosetta. I followed the codes of the paper below (which is mostly similar to Rosetta ligand docking) to design the ligand-binding site. I generated 2000 outputs, but only around 200 designs have interface scores less than WT, and the rest (1800 designs) have binding scores greater than the one of WT (even positive binding score). I know Rosetta might keep designs that are slightly worse than the best design to ensure it does not get stuck in the local minimum. But I think something might be wrong with my simulation (based on the number of good designs). Do you have any suggestions to fix it? Or do you have any explanation about it?
I ran designs for different poses. In one of them, the problem is more significant, and less than 10 outputs have better scores than WT.
I set the bonus for favor_native as 1 (as the paper). Can that have a negative effect?
Also, I wanted to know whether it is possible to design the ligand binding site with Rosetta such that it can improve binding with two ligands at the same time?
Thank you
P.S: I performed one round of design, but I think even one round of design it should not generate poses with interface score worse than than WT.
.
