[gmx-users] generating tripeptides in a box

[Pim Frederix]

Hello users,

I'm experiencing trouble using genbox for creating a box of tripeptides in MARTINI CG polarizable water. I feel like any of the options below should do the trick, but all give (different) problems. 

Generating the randomly place tripeptide box works fine, nothing unusual about the output:

genbox -ci tripeptide.pdb -box 12.5 12.5 12.5 -vdwd 0.3 -nmol 300 -o tripeptide_box.gro

> In which tripeptide.pdb is the coarse-grain coordinates generated either with Martini 2.1 or Martini 2.2P (martinize.py)

But then trouble starts.
I have got the polarizable water from the Martini website (also normal, single bead water gives problems in method 1&2, seems to work ok for method 3)

1) genbox -cp tripeptide_box.gro -cs polarize-water.gro -vdwd 0.21 -o tripeptide_water.gro     (vdwd 0.21 is needed for Martini water)
This gives me a result that looked fine at first, until I realised that somehow a simulation box that has a greater density of water molecules in the centre of the box. It's hard to describe, but it looks like a cubic water box with high density in the middle of a normal cubic water box. There's no gradual gradient, it stops at a certain distance. Hopefully this ASCII art side view will help:
_____
|   __   |   
|  |__|  |
|_____| 

If I try to minimize/equilibrate this box it just explodes, as the density in the center is about twice the appropriate density.

2) I though this may be because the water box is not large enough (it's about 5x8x5 nm, my system 12.5x12.5x12.5). However, firstly generating a solvent box that is large enough to cover my whole tripeptide box, using genconf, doesn't help and gets me back to the same problem as in 1)

3) I discovered that with the single-bead water it works if I make a water box that is exactly the size of the tripeptide box, the problem is minute and gets annihilated by just minimizing and equilibrating the box. However, if I do this with the polarizable water (3 'beads' per water molecule) this doesn't work anymore. I get my solvated box with the right density, but there are a few water molecules (it varied from about 2 to 10 in a small number of attempts)  that are completely out of the box some 10 nm away. Manually removing them is not an option unfortunately as I need to automate this process for a large number of peptides. 
Also, if I get the number of waters from the residue information and put in the topology file, the number of coordinates in the .gro file doesn't match the number of coordinates in the .top, there's a varying mismatch of around 100 atoms (out of ~17000).


I've posted this problem about 2 years ago as well, but didn't get any replies except from people that wanted to know if I solved the problem already. So I really hope someone can make sense of this and tell me where it's going wrong (either genbox (which I think) or something to do with the Martini molecules) and can suggest something to solve it. I'm happy to send files if needed.

Many thanks,

Pim F.
University of Strathclyde
Glasgow, UK