Repository navigation
Too many bonds detected #1
Description
I was testing the code for a CHARMM36 simulation from here:
https://doi.org/10.5281/zenodo.3474862
and it seems that sometimes the code detects too many bonds. Here is the detailed description of what I did:
I made a pdb file of single POPC with make_ndx and trjconv selecting the first residue
gmx make_ndx -f mdrun.tpr (selection r1)
gmx trjconv -s mdrun.tpr -f md_in.gro -o POPC.pdb -n index.ndx
and gave the resulting POPC.pdb as an input to the code:
python3.7 /home/samuli/work/NMRlipids/autoLipMap/autoLipMap.py -p POPC.pdb -l POPC -om mapping.txt -od defition.def
I get an error "Actual POPC has 135 bonds, while it should have 133"
It seems to arise, for example, from the detected bond between C28 and H7R attached to C27: "Add bond #61 between atom 58 (H7R) and atom 60 (C28)"
Activity
patrickfuchs commentedon May 15, 2020
Thanks @ohsOllila ! I think the problem comes from the graph generation when it infers bonds (edges) from the PDB file. Basically, there's a double nested loops on all atom pairs, then it calculates a distance between each atom pair and decides whether there's a bond if this dist is below the constant named
MAXBOND. This constant is at 1.8 (A). If I set it to 1.7 it works with the PDB you tried.A typical C-C length is 154 pm, so 1.8 A was too long. Initially I set it to 1.8 because I thought in some cases some bonds might be stretched. But 1.7 is probably better. Do you think I should reduce it more?