GROMACS modification: No
I have run alchemical transformation to study the stability of proteins as effect of mutation by calculating the delta delta G (dd G) using REST2/FEP technique.
To do so, I prepared dual .top file first which contains the info. of the mutant residue in addition to the wild type residues.
In this case, first replica the wild type form while the last replica is the same structure with one mutant residue.
Now I am looking for applying the same method to study the effect (based on the dd G value ) of stability of the proteins due to insertion or delation of one or two resides.
I looked for any article about using this based on alchemical transformation I couldn’t find!!
Anyway, if you have any idea how to prepare the top file in GROMACS to be able to run REST with. different structures to get first replica the wild type form while the last replica should be the same structure after deletion or inserted the residue then I can use BAR(FEP) to get the dd G between these two forms
Unless you want to insert or delete the residues at or very close to one of the termini, this can become a very tricky problem. Though I never tried to run such simulations myself, I see two approaches you may try, which both come with significant disadvantages, unfortunately:
Very likely to work, but probably a performance nightmare.
From the insertion/deletion point towards the closer terminus, you include all atoms in the FEP region such that you can alchemically transform all these amino acids into the amino acids they are in the insertion/deletion mutant. This can in principle be done with the very same protocol as you used before since you encode the insertion/deletion as a series of point mutations. However, the FEP region becomes very large and the simulations thus very slow.
Same performance as a single point mutation, but is likely to have convergence issues/spurious free energy contributions from the large-scale protein movements required.
You just turn the residue you want to insert/delete into a residue of dummy atoms in the respective state. (It’s technically just a single point mutation like in your previous setup.) Moreover, you identify the bonds, angles and dihedrals that are formed once the residue to be deleted is gone. These bonded interactions get a force constant of zero as long as the residue to be deleted is still around and are turned on once the residue to be deleted has been annihilated. This solution is computationally much cheeper, but you will have to play around with the lambda vector settings in the MDP file. In particular, consider to fully convert Coulomb and van der Waals interactions before changing the bonded terms to avoid forcing neighbouring residues to crash into the residue to be deleted. And if you manage to get this going, this approach requires both parts of the protein to move to fill the gap left by the residue to be deleted. Therefore, I’d assume that this protocol can only work for an isolated protein in aqueous solution. And it might be needed to let the dummy atoms of the backbone of the residue to be deleted still have repulsive vdW interactions with the water molecules to prevent them from moving between the two protein parts as you annihilate the residue. When trying this approach, I urgently recommend consistency checks like: Do I get the same free energy if start the simulation from the wildtype and from the insertion/deletion mutant? It really has to be validated that you don’t get spurious free energy contributions from forcing the two parts of the protein to move to fill the gap left by the residue to be deleted.