GROMACS version: 2024
GROMACS modification: No
I am running some hbond analysis because I want to take a look at the interactions between an arginine (ARG) in a protein and a lipid bilayer with two lipid types, cardiolipin (TLCL) and POPC. I am using GROMACS2024 and the tool gmx hbond and I do not modify the standard selection criteria within the tool, the only flags I use are for input/output.
Now, what happens is that I select the group ‘MEMB’, which contains both POPC and TLCL, and I get my hbond num vs time data. Then I do the same but specifying POPC and TLCL to check which of the two is contributing the most. TLCL works, while for POPC I get
Inconsistency in user input:
Selection ‘POPC’ has no donors AND has no acceptors! Nothing to be done.
Now, same files, but I use gmx hbond-legacy and…it works! I get the number of h-bonds vs time in all three cases, TLCL-ARG, POPC-ARG, and MEMB-ARG.
What is more puzzling, though, is the result. TLCL bounds very rarely with ARG, and this is shown with consistent results by both tools
For POPC I clearly do not have the corresponding non-legacy version for the tool as per the latter, and I get this with legacy
However, when I compute the h-bonding of ARG with the whole bilayer, I obtain the following
Basically, the new tool applied on the whole membrane sees all the hbonds, a;so the POPC ones, as it can be seen by comparing POPC//legacy with MEMB//new. The only discrepancy is the peak around 286ns where the contribution of TLCL is missing in the POPC analysis, as it should be.
I thought this could have been an error of mine in the groups definitions, but I
i) visualize with VMD and clearly see hbonds with the same criteria between ARG and POPC
ii) I am using the same files with the two tools (the same index file in particular)
iii) I get the same results as gmx hbond-legacy if I use the normal gmx hbond tool in GROMACS2023.2
I tried to read the documentation for both hbond-legacy and hbond in the 2024 version, but I can’t see what the cause can be. I also tried to specify in the opposite order the group indexes (e.g. provide MEMB and then ARG or ARG and then MEMB) to the tools but they give the same results (as it should be), so I guess it’s not related to the parsing of the indexes.
EDIT: may be related strictly to this issue.