Re: Question

From: McGuire, Kelly (mcg05004_at_byui.edu)
Date: Thu May 02 2019 - 14:04:31 CDT

Thanks for the guidance Brian. I haven't successfully finished a string method followed by the eABF for a QMMM simulation. My mentor is under the impression it would take too long to see any results. In your opinion/experience, is the string method and eABF faster than we are imagining for QMMM simulations? In MM ABF, I get plenty of sampling in each window because I can get 45 nanoseconds per day per window. How does eABF obtain enough samples when QMMM is on the picoseconds timescale? Thanks!

Kelly L. McGuire

PhD Candidate

Biophysics

Department of Physiology and Developmental Biology

Brigham Young University

LSB 3050

Provo, UT 84602

________________________________
From: Brian Radak <brian.radak_at_gmail.com>
Sent: Thursday, May 2, 2019 12:59 PM
To: namd-l; McGuire, Kelly
Subject: Re: namd-l: Question

Hi Kelly,

I recall seeing something like this (using umbrella sampling) proposed in JCTC several years back - definitely not a standard kind of calculation. There are intrinsic difficulties in "stitching together" QM and MM results since the difference in zero of energy is often hard to define (and depends on the method). Maybe the fact that ABF relies on force helps?

In short, you can probably run the simulations just fine but you might be all alone in trying to figure out how to analyze the results.

HTH,
BKR

On Thu, May 2, 2019 at 1:12 PM McGuire, Kelly <mcg05004_at_byui.edu<mailto:mcg05004_at_byui.edu>> wrote:
So, I was just asked if it is possible to mix QM and MM with ABF. I know QMMM is possible with the string method and then using eABF. I am working on the tutorial for that. But this question was a bit different. Here is the example:

One of the windows has copper in the protein binding site. So, for that window, use QM. Then, in every other window along the protein reaction coordinate, treat the copper with MM parameters. Only do MM for the rest of the reaction coordinate. He was basically asking, can you tie in one window that is treated with QM, to all of the other windows treated with MM, and get a free energy profile? Or, as I move the copper along the reaction coordinate, do those windows need to be treated with QM so that all of the windows can be merged properly? He is worried that QMMM string method/eABF will take to long to get a free energy profile for our copper complexes in the protein. But, we can't completely trust any MM parameters that exist for copper...With MM ABF, I can get ~45 nanoseconds per day in each window. But with QM/MM, I can only get ~1.5-2 picoseconds per day with 4 GPUs using Terachem. Which is ~32x faster than ORCA without GPUs. But, how is it possible to get enough sampling for eABF in each window with QMMM?

Kelly L. McGuire

PhD Candidate

Biophysics

Department of Physiology and Developmental Biology

Brigham Young University

LSB 3050

Provo, UT 84602

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2019 - 23:20:45 CST