[gmx-users] umbrella sampling part of a molecule with constraints=h-bonds

[chris.neale at utoronto.ca]

While doing umbrella sampling, can I use constraints = h-bonds and  
then select groups that are a only portions of a molecule, but are  
composed of whole amino acids such that the groups are joined to the  
rest of their molecule by non-Hbonds?

For example, pull on a group that is composed of residues 2 and 3 but  
not 1 or 4 ? As detailed below, the group would not be "connected by  
constraints to the rest of the system". However, the next sentance in  
the manual refering to limitations reports that "if a group contains  
part of a molecule of which the bondlengths are constrained...results  
unreliable." The explicit reference to constraints=all-bonds in the  
manual makes me think that what I am suggesting is actually ok. If so,  
the manual could benifit from some comment that constraints=h-bonds  
would allow a user to define groups as parts of molecules as long as a  
H-bond does not straddle the group boundary.


     Umbrella only on these 2 boxed residues ??
            --------------------------
            |                        |
            |                        |
            |        O               |         O
            |        ||              |         ||
    CA      | N      C       CA      |N        C
  /    \   /|   \  /   \   /    \   /|  \    /
N       C  |     CA     N        C  |    CA
        ||  |                     || |
        O   |                     O  |
            |                        |
            --------------------------


The entire relevant passage from the manual reads:
Quoting Limitations of umbrella sampling Manual3.3 section 6.2.5
"constraint forces can only be calculated between groups that are not  
connected by constraints to the rest of the system. If a group  
contains part of a molecule of which the bondlengths are constrained,  
SHAKE or LINCS and the constraint force calculation here will  
interfere with each other, making the results unreliable. This means  
that for simulations with constraints = all-bonds in the .mdp file  
pulling is limited to whole molecules or groups of molecules."