Re: [AMBER] TI residue mutation with pmemd on GPU

From: David Case <>
Date: Fri, 20 Sep 2019 12:56:16 +0000

On Fri, Sep 20, 2019, Hosein Geraili Daronkola wrote:

>I am trying to mutate ASP to ASN on a protein using the TI one-step method.

First: be sure that you have an updated version of Amber: see for more information.

>I use "alchemical_analysis" to get free energy, and because the whole
>difference that I wanna show is around 1 kcal, I need to have a low
>standard deviation.

I'm assuming you mean here the difference (say) between mutating ASP to
ASN in the protein, and in solution (or some other similar free energy
cycle.) The magnitude of the free energy change along any single let of
the cycle is not physically meaningful.

>The problem is with increasing the number of lambda to
>45 or increasing the simulation time, I never reach to a result with a TI
>dhdl graph without a sharp kink,

At what value of lambda do you see the kink? Generally, it should not
be necessary to use large numbers lambda points to carry out a nearly
isosteric change like this one. Note that the TI analysis itself
doesn't depend on phase-space overlap between neighboring lambda points.
Try an initial simulation with 7-10 lambda points, and see if you can
get <dV/dlambda> curves that make sense.

Maybe someone else on the list will spot a problem with the details of
your input.

...good luck....dac

AMBER mailing list
Received on Fri Sep 20 2019 - 06:00:06 PDT
Custom Search