Do I really need a "good" PDB as a starting point for pdb2gmx?

GROMACS version: 2021
GROMACS modification: /No
Here post your question

I use several software to do homology modelling, including Rosetta, MODELLER, SWISS-MODEL, I-TASSER, etc. The online servers SWISS-MODEL and I-TASSER could give less than 10 best models. The local software Rosetta and MODELLER are required to generate hundreds or thousands of models, and choose the best one. So it would only take 2-3 days to get the results from SWISS-MODEL and I-TASSER, while it could take weeks to obtain thousands of models from Rosetta and MODELLER.

I have heard the saying of “Rubbish in, rubbish out”. So it is important to make sure the starting PDB is good enough. But is it really necessary?

The Gromacs already has a step of energy minimisation, so it seems that any suboptimal bonds can always be rearranged to the desired states. Is that correct?
gmx grompp -f minim.mdp -c protein_solv_ions.gro -p topol.top -o em.tpr
gerun mdrun_mpi -v -deffnm em

Energy minimization doesn’t correct for entirely poorly predicted geometries, bad loops, etc. It only refines the coordinates provided and does limited correction. You can get a perfectly sensible energy minimization on a molecular ball of spaghetti (like I’ve seen come from some programs…) but it doesn’t mean a resulting simulation will have any meaning.

1 Like

Thank you. So it is worthwhile to prepare an initial structure good enough for pdb2gmx.