-
Notifications
You must be signed in to change notification settings - Fork 1
Placement
I use Fragmenstein for placement. Fragmenstein can place compounds in 2 ways.
- Using PyRosetta for minimization of compound overlaid on atom mapped atoms to fragments. Called the Victor class. Find the code here.
- Does NOT use PyRosetta for minimization in protein. Minimizes conformation using Merck Molecular Force Field (MMFF). Then puts this minimized conformation into fixed protein template. Called the Wictor class. Find the code here.
As of 7 Nov, I have been using method 1, with PyRosetta. Now I am reconsidering. The elaborated compounds will not be significantly different than the fragments or in very different locations. Residues involved in interactions will adopt similar conformations as in the template structure I would use for placement anyways. Therefore I think placing the pre-minimized conformation with MMFF and forcing into pocket will work okay. I will need to test this to prove.
I will have to update this section once I get more concrete information. After chatting with Matteo on Tue 7 Nov, PyRosetta minimizes the ligand and centroids of residues within 8 angstroms of the centroid of the ligand. The centroid of the ligand is determined by ???. The centroids of residues is the beta carbon.
PyRosetta uses a hybrid FF to model waters, but does not account for rigidification or desolvation. In the final output score of Fragmenstein Matteo weights rigidification. According to him "rigidication comes at a cost of 1-2 kcal/mol, so I use a penalty of 1 kcal/mol". A molecule with 5 rotatable bonds has a penalty of 5 kcal/mol which is significant.