I am trying to use RosettaLigand to dock a rod-shaped molecule into a protein binder with a deep, narrow binding pocket. Due to some experimental limitations, the ligand can be bound only in specific orientation - a ligand edge that is conjugated to an attached large molecule should be headed to the protein surface.
I have placed the ligand in the right orientation at the starting point. I don't want to use constraint file with random distances' constraints as is could be limiting and problematic, so I tried to limit the rmsd in the transform mover; to not use the initial_perturb option (or to use it only with small-angle perturbation); to limit the translations and orientations in the HighResDocker mover - and I still get flipped orientation. I also tried using the DiffAtomBurial filter but inspecting manually the results, it doesn't seem to work as expected. I then moved to use translate-rotate-slide together because of the ability to limit the degrees in the rotate mover. However, even limiting the degrees in rotate to 30 results in flipped orientations (even if I stop the protocol after slide together, meaning with no further high-res docking and minimization). The logs suggest that only the phi and psi but not theta are indeed limited. I am not quite sure how the axes of rotations are defined in relative to the ligand, but I guess they are not defined well with respect to the specific orientation I wish to maintain. Is there a way to influence this? Or maybe some other way to tackle this specific-orientation problem (without using cstfile and without severe restrictions on the movements in the "correct" axes, as the binding pocket is quite deep)?
Are you suppling rotamers for the params file? These get used as shortcuts in some steps, like repacking I think. So maybe removing the rotamer line from your topology file will fix the bad behaviour?
Also, given your ligand I am guessing it might be best switching from internal coordinates to cartesian.
However, for covalent attachments you will need a constraint file to make sure the attachment is not bent unnaturally —one distance and two angles and maybe one dihedral at stregth 5 will fix it.