Bruno Kestemont Lv 1
As I see that I'm curently top soloist for small molecule design, it's time to share my method for inspiration of other players and for Science. Don't forget to share your good solutions to Scientists (good means an almost maximum Bonus and several blue HBonds to the target). Please feel free to share your strategy in the comments. We have still a lot to learn one from the other.
Here is my default strategy for a solution that the scientists found interesting for an old puzzle:
It's mainly a random try and evaluate strategy. My focus was all the time to maximize the objectives. I show the unsatisfied objectives and I try to solve them by deleting or changing several atoms or liaisons between atoms (I follow the tips shown on Mouseover). I don't look at the HBond or rotation objectives because this often solves automatically when wiggling afterwards. When the Foldit team proposes us a very small Ligand at start, the way is to add atoms or structures. When they start with a very big molecule, the way is deleting a lot of atoms and structures, trying to keep the existing blue HBonds.
I might add some structure pointing either to voids or to polar atoms of the receptor. I almost only tried the same simple structures: hexagonal aromatic with only one blue polar atom, "8" aromatic with one blue polar atom depending on the void available or proximity to polar atom, or a simple "Two atom in line" structure. For the common structure linking blue and red atoms, I deleted or changed both to C, or replaced the N on the other side.
When almost all objectives are met (unless Hbond and rotations), wiggle and move. When all objectives are met (including HBond), load this ligand to Library for evaluation.
If Library doesn't propose ligands, delete or add several atoms or structures and try again untill Library proposes something.
If Library proposes something good, and full bonus is reached after download and wiggle, we got a good starting point (see video). It's important to have at least one blue hydrogen bond (move the ligand if needed to achieve this condition, of add bands between donors and receptors, between red and blue, with length <2).
Then when we are finished with hand fold, it's time to use recipes overnight and go to sleep or work, eat etc.
Recipe FBS 2.3.1 no mutate overnight. This tries several random ligand displacements always starting from the same position. Then Ligand Docker and Rav3n_pl GAB v3.4 filter Cuts LIGAND, finishing with Medium wiggle power. Sharing with scientists if the full bonus is still present.
Then trying to replace, delete or adding C here and there, wiggle and recipe Ligand Docker in order to rise the score.
Then taking the best result and either starting again from the * above, either reset the puzzle and starting from scratch or from a similar library proposal (building on previous experience with this particular puzzle).
