Solvating without having water molecule "inside" the solute

GROMACS version:
GROMACS modification: No

I am trying to solvate a bunch of cellulose chains with waters using gmx solvate. but it seems that no matter what I do, there will be water molecules (shown in yellow) “in between” the cellulose chains (shown in cyan) since the chains are sufficiently far away from each other. Is there a good way for me to force the solvation to happen only “outside” of the entire packed cellulose structure (outside the outmost cellulose chains)?

Or equivalently, is there a way for me to specify a minimum distance between solute and solvent during solvation without messing up the water density (meaning not using gmx solvate -scale)? If so, I can simply define a larger value for this minimum distance.
image
Thanks in advance!

Hi, have you tried changing -radius?

Hi Thanks! From GROMACS manual,

. A database (vdwradii.dat ) of van der Waals radii is read by the program, and the resulting radii scaled by -scale . If radii are not found in the database, those atoms are assigned the (pre-scaled) distance -radius .

This seems to suggest that -radius only has an effect when the vdW radii of the atom is not found in the database, but my atoms are just C, H, and O which are all in the database already. Did I misunderstand something over here?

Ok… Then I guess you can remove water afterwards by doing a selection (gmx select) of cellulose + water outside the fibers.
https://manual.gromacs.org/documentation/5.1/onlinehelp/selections.html