GROMACS version: 2018.3
GROMACS modification: No
Hello everyone,
Have a newbie questions:
(1) Why most of Martini tutorials/protocols for proteins suggest using RF electrostatic treatment? Which is essentially accounting for very closest water molecules explicitly and replacing the rest with dielectric constant.
I have tried modeling and simulating in Gromacs large protein-water-ions systems with both RF and PME treatments (no lipids). In PME case I’ve observed water “freezing” - sudden drop in wat-wat LJ-short-range interactions energy at 310K after ~100ns of simulations.
(2) Is this related to RAM distribution between cluster nodes and huge rounding errors?
(3) The papers I found on this matter mention using 10% of antifreeze beads to avoid water freezing and producing some sorts of phase diagrams. Is there a way of fixing it by tweaking FF parameters without introducing anti-freeze beads??
(4) Another interesting effect was salt ions sticking to each other when simulated in pure water or sticking to protein/other ions. Although I saw some ions aggregation in 1 microsecond all-atom simulations before, it wasn’t as pronounced. Is it an artefact of Martini FF?
(5) In RF simulations I didn’t see any of the “freezing” or ion aggregation, but protein Rgyr is extremely reduced even with recommended correction to LJ prot-wat interactions of 1.1-1.2. I don’t see such collapse of protein in PME. One of the paper on dipeptides aggregtion concluded that Martini3 does not allow for proper protein aggregation/formation of liquid phase droplets, while Martini 2.1 with tweaking LJ parameters was applicable.
Would appreciate any comments/suggestions. Thank you.