normal Inquiry regarding coarse-grained forcefield for non-bio molecules.

  • sonics01
  • sonics01's Avatar Topic Author
  • Offline
  • Fresh Boarder
1 year 5 months ago #7552 by sonics01

We are looking for a Coarse-Grained model to simulate the mixing / diffusion of:

a) GPS (glycid-oxi-propyl-trimetil-oxi-silane, C6H14O5Si)

b) Jeffamine molecule (C9H22O2N2)

Our system has 5 atom types (C, H, O, N, Si) and 2 types of molecules (GPS and Jeffamine). We wish to run huge scale simulation (at least 500,000 or more atoms) of this system for very long time, like hundreds of nano seconds order.

We tried conventional MD simulation, but because of scale limitation, it takes too long time to simulate such system using empirical atomistic force field methods. That is why we are looking for Coarse-Grained model. Our target is to run the mixing/diffusion simulation of two molecules, and check how the diffusion progress looks like w.r.t. time.

Are there any Martini CG force field that we can apply for our system?

If there are no such CG force field, then approximately how long would it take to train Martini CG force field for our model?

Thank you

Best wishes,

Please Log in or Create an account to join the conversation.

1 year 5 months ago #7571 by Pim
As far as I know your molecules have never been studied using the Martini force field, and there are some non-standard groups (like silane and the epoxide). Sadly, that means there is no easy solution for your problem. The good news is that you already have your atomistic data as a reference. That speed things up in terms of the development of a CG Martini model. Depending on your desired level of accuracy, it would still take a few weeks to make a reliable model I'd say. Check out the tutorials "Parametrizing a new molecule" and "Polymers", together they should give you a pretty accurate picture of what would be necessary.

If I understand your question correctly, you want to study a sort of melt of the two molecules, or are they dispersed in solvent? In any case, you will indeed need quite some simulation time if you want to study the diffusion. You can probably get the speed-up you need with Martini compared to all-atom ff's, A 500,000 atom system would not be such a problem on the CG level to simulate for a few microseconds, I would expect the simulations to be ~100 times faster than .your atomistic simulations.

Please Log in or Create an account to join the conversation.

Time to create page: 0.139 seconds