Pardon the lengthy post…i hope to give a better picture so that i could receive better advice on how to proceed.
I would like to study the interactions between small molecules and a protein.
In this set up, i imagine the most efficient way would be to have 1 protein in the middle of the box, surrounded by water and another small molecule. To observe meaningful interactions between the small molecule and protein, i would like to have multiple small molecules in the system (approximately 20-30?). Small molecules would be randomly positioned.
As this protein is pretty large, i want to minimize the computation time and to impose constraints on the system but im very new to performing such customization of the system.
Ive only had experience with setting up proteins and small molecules in a solvent and letting them run freely.
First of all, let me describe what i feel is important in this system.
The secondary/tertiary structure of protein is important and i want the structure to be the same at the end of mdrun.
I want to have the small molecules to be unrestrained because the molecule conformation might change upon interacting with the protein. But i can consider adding full restraints to small molecules if it speeds up computation time.
Does gromacs have the function to fix the protein position in solvent box while allowing water and small molecules to move freely around it? Which command lines are able to control this?
From what i have observed from previous mdruns for investigating interaction between multiple molecules, there seems to be not much movement overall (i had imagined molecules to move towards each other, bump, and move apart, and bump other molecules…like brownian motion). But what i observed is that molecules that are close to each other will interact and remain pretty stable, while those further apart remain apart after mdrun.
If i were to mimic the system experimentally, the molecules might be spaced too far apart in solvent box to interact within a sufficient time frame.
Is it a common practice to ‘increase’ the molecule concentration (decreasing separation distance between molecules) in order to observe interactions better?
Thank you in advance for any advice, and for taking the time to read through this.
This is what position restraints are for. They’re already in the topology. Use define = -DPOSRES in your .mdp file.
You’re likely going to suffer from what limits most MD simulations - getting stuck in a local minimum.
What you’re seeking to do sounds a lot like SILCS sampling from the MacKerell group. Coupling Grand Canonical Monte Carlo sampling overcomes the problem of molecules getting stuck, e.g. dx.doi.org/10.1021/ct500201y
Whether you can come up with a similar protocol or not is another challenge. SILCS is designed for pre-programmed fragments, and is not going to be something you can just drop your molecules into and run. But you should seriously consider enhanced sampling strategies like GCMC (unfortunately though, GROMACS cannot do this natively).
Sorry my mistake, i meant to increase sampling time. In a way, to speed up simulation for same overall time required, but yeah i got what you mean.
Thanks! I just managed to get a better idea of how this works. Just to clarify, sometimes for ligands or other small molecules, their restraints in .itp file are labelled as POSRES_LIGAND. Does this mean i have to follow this naming in .mdp file and label it as define = -DPOSRES_LIGAND ?
This does sound immensely complicated, especially for myself as ive barely understood the functions of gromacs. Are there alternative methods to this? I have come across the webSDA server and it has a multiple molecules function, mainly to observe clustering and possibly interaction sites. Sadly, the webserver only allows single-type molecule for this function. I assume this function also treats the molecules as rigid bodies. By rigid body, is this similar to constraining all bonds?