Re: [AMBER] Does minimization require for average pdb structure extract from 2drms plot using CPPTRAJ AMBER 16

From: Thomas Cheatham <>
Date: Mon, 25 Oct 2021 20:21:11 +0000

It is a matter of perspective. Yes, the examples so far are valid (methyl groups, sine wave) - another good one is the average structure of cyclohexane (which is flat and does not represent the two distinct, but chemically equivalent, chair conformations). However what you see experimentally depends on the temperature. At low temperature, experiment may show the different conformational states (since they get trapped long enough to been "seen"). However, as the temperature goes up, you invariably are looking at the average... For example, in proteins, you normally do not see the separate cis/trans of the peptide bond (noting that trans dominates) but with proline, the barrier to interconversion is high (~20 kcal/mol) so you capture both states in experiment. With cyclohexane at room temperature, 1H NMR shows a single peak; as the temperature lowers (< 100 deg) this separates into two peaks.

In our work on force field validation, we have shown we can get Dickerson dodecamer structures within 0.5 Angstrom of the average NMR 1NAJ structures when comparing the average structure determined from the MD (in long simulations). If you compare instantaneous snapshots from the trajectory (that are physically "correct"), these are 1-1.5 Angstrom from the average structure from experiment. So depends on what you are trying to look at and compare to ... for docking post-facto to MD simulations, minimizing the average (or top cluster populated averages) structure makes sense as does selecting a series of either random or representative snapshots.

In a final note, regarding chemically equivalent states of a molecule, if not addressed, can lead to artificially higher RMSD values - an example is phenylalanine where the side chain flipped state (rotated 180 deg) is chemically equivalent, but atom names do not match so higher RMSD. CPPTRAJ has the capability to correct for this...


AMBER mailing list
Received on Mon Oct 25 2021 - 13:30:02 PDT
Custom Search