I haven’t used PACKMOL. Could you increase the tolerance to 6.0 (or something) and see if it helps?
Actually, using a higher tolerance on PACKMOL would lead to large gaps between different regions through which water would enter into the bilayer. The general treatment is to use a tolerance of 2.0 Angstroms to avoid that.
Then I’m afraid we’ll have to see if someone else can help you.
I see that you already tried many different things and your minimization protocol looks quite solid. Have you tried visualizing what is happening during the equilibration? Do the warnings come from a specific region in your membrane (like, always the same lipid molecule) or does it change once you set a different distribution of velocities? If you have just a few troublesome lipids you could try removing them from your system and adjusting the topology accordingly. Also, is it possible for you to / have you tried MARTINI 3?
Dear fatpmeireles,
Thank you for your input. I have not tried using MARTINI 3.0 yet. I have some how managed, by sheer dumb luck and black magic, to get the equilibration runs to work for two systems, the ones marked 0.5_LPE and 5_LPE. I reran the minimization step for a slightly different value of emtol which seemed to do the job and did not lead to any LINCS errors, or any errors for that matter. The same procedure have not been fruitful for the other two systems.
As far as the lipids go, they seem to be cholesterol molecules at the edges of differentially packed regions in the membrane. Removing them may work, but it would mess with the stoichiometry of the systems which would be less than ideal.
I see! It is likely a weird packing issue if it worked for the other two. I don’t know if you already solved it at this point, but what you could try to do is translate these cholesterol molecules to the solvent, still quite close to the interface, and let them diffuse back to the membrane. I assume this would let them pack back more easily.
Dear MagnusL and fatpmeireles,
I have finally managed to fix all of the issues. Turns out while defining constraints on packmol for the different lipids, I had set the restriction for the cholesterol molecules in such a manner that the cholesterol ROH groups were exactly in line with the lipid head groups. In reality, they lie slightly below the lipid head groups. Making this change almost immediately fixed the problems. Still multiple minimization runs were required to get the system stable enough for equilibration runs.
Great to hear that you got it to work and thanks for your explanation. Hopefully it will help others.
Dear @abelxf ,
Glad to see you were able to fix your problem. Something else that you might consider are the rcoulomb and rvdw values in your .mdp file. These seem to be set at 2 nm, which deviates from the standard MARTINI values. Not sure if this is intentional or not.
Dear adq,
I actually started out with the default values of 1.1 nm. During the very first run of minimisation, I got an error which was along the lines of
“There n number of non-perturbed interactions …”
I followed the CHARMM GUI prescribed minimization/equilibriation steps. CHARMM GUI provided readme files ask you to change these cutoffs to 2 nm, if this particular issue is encountered. The exact physical reason as to why we do so I am not aware of.