Okay, it is clear now.
Hello all,
One could refer to this refer to this paper https://pubs.acs.org/doi/10.1021/acs.jctc.0c01328 to get a clear view regarding dummy atoms in free energy calculations especially in ligands.
This is not the method currently implemented in pmx
In that case, I wanted to ask about atom mapping. Just wanted to understand about the mapping as I have seen instances in atom mapping when I use the pairs2.dat file to build the “ligandHybrid”(pmx ligandHybrid), I have encountered errors.
The conclusion in the paper suggests of “adjustments of force field terms involving dummy atoms” which I understood is the case with scaling of dummy masses, angles and parameters.
The paper offers a specific set of rules, pmx uses a different decoupling approach.
1 Like
pairs2.dat is the same mapping, but ligandB is considered first, ligandA second, i.e. the atoms are mapped:
atomK_ligandB atomL_ligandA
atomX_ligabdB atomY_ligandA
…
1 Like
Does this scaling have a significant impact on the ddG values? (as I have observed in my case, just wanted to know whether what I have observed is an anomaly or it is the usual case)
I am calculating RBFE of ligands.
In my experience, scaling of those interactions has only minimal effect on ddG (dGs of the legs in thermodynamic cycle will of course be very different). How large is the effect you see?
About 4 times.
Is this outside the range of uncertainty? What were the exact dG values in this case?
For one of the arm, in one instance it is 0.54 KJ/mol and in another instance(scaled-down) it is -15.94 KJ/mol. It seems pretty strange.
Another thing I wanted to ask is that do you use equilibrium RBFE of these pmx derived ligand hybrid topologies?
Does then the other leg of the thermodynamic cycle change asymmetrically?
You can use any alchemical approach with the hybrid topologies.
Yes for the other legs the numbers are 37KJ/mol and for the scaled-down it is -24.03KJ/mol.
Are these values converged?
Haven’t done multiple runs. Just speaking from a single run.
What uncertainties are reported for the single run? What is the convergence value in the output? Are the work distributions overlapping?
Yes they are overlapping.
I do not get this.
You mean to say the standard deviations?
Please, paste the results files for each of the runs you describe