Re: [AMBER] QM/MM adaptive biased MD_ABMD

From: Aixiao Li <>
Date: Fri, 28 Nov 2014 21:32:36 +0100 (CET)

 To make things clear, I take the amber tutorial as example:, here the tutorial used the ncsu_smd which is also adpative biased. What I did is just replace the ncsu_smd by ncsu_abmd. The free energy study is one dimensional, with the distance between atom 8&9 decrease from 1.74 to 1.04. and the concerning input part is below. But when I check the abmd.txt, the values stay around 1.7~ 2.0A.
 And I still have the question, anybody knows how to define kernel width for ABMD?

timescale=200.0 !not so clear, in ps

   type = DISTANCE ! Linear Combination of Distance
   i = (8,9)
   min=1.04 max=1.74
   end variable

end ncsu_abmd

----- Mail original -----
De: "David A Case" <>
À: "AMBER Mailing List" <>
Envoyé: Vendredi 28 Novembre 2014 18:28:25
Objet: Re: [AMBER] QM/MM adaptive biased MD_ABMD

On Fri, Nov 28, 2014, Aixiao Li wrote:

> I am currently running QM/MM ncsu_abmd for free energy calculation. It
> is just a simple proton transfer from water to O atom (of GTP).

Have you done a simple umbrella sampling calculation to get a rough free
energy surface, and hence to have confidence that your QM/MM model is a
reasonable one, and that your chosen progress variable(s) indeed span the
reaction? These are items that generally should come before running ABMD
calculations. Your ABMD results (you never get to small values of the proton
transfer coordinate) might be the result of a poorly-chosen Hamiltonian:
proton-transfer events can be very hard things to study.

...good luck....dac

AMBER mailing list

AMBER mailing list
Received on Fri Nov 28 2014 - 13:30:03 PST
Custom Search