AMBER: Statistical uncertainty estimation in Thermodynamic integration calculation..

From: cgji <>
Date: Mon, 24 Sep 2007 21:13:36 +0800

Dear Amber users,

 Recently I tried to use molecular dynamics free energy simulations (MDFE) to compute proton pka
 shifts as described in amber tutorial (
This tutorial reproduces the calculation of the pKa value of the ASP residue in the protein thioredoxin as described in the following paper:
Simonson, T., Carlsson, J., Case, D.A., "Proton Binding to Proteins: pKa Calculations with Explicit and Implicit Solvent Models", JACS 2004, 126, pp4167-4180.
 In table 1 of the article ,they showed us expilicit solvent MDFE results for Thioredoxin ASP26 .

 Part of this table was listed bellow ,


 lamba=0.11270 -3.1(12.0)
 lamba=0.50000 -64.5(4.6)
 lamba=0.88279 -131.4(7.8)

 Delta G -66.0(3.9)

 Delta,Delta G 9.1 (4.1)

 12.0 ,4.6 and 7.8 in parentheses was statistical uncertainty for the three
 seperated trajectory .
My question is : can I estimate statistical uncertainty from each seperated trajectory ?
in their article , they said "for the derivatives , it is estimated as twice the standard derivation of block averages ,each trajectory being divided into four blocks." , however , i can't understand it clearly .
2. How can I use these data from seperated trajectory to construct statistical uncertainty (3.9) for
Delta G ?
 I tried to calculate it as follows

 sqrt ( (12.0*0.27777)**2+(4.6*0.44444)**2+(7.8*0.27777) **2)
 but I can't get 3.9 .

I appreciate it very much for your help......

Yours sincerely,
Changge Ji


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Mon Sep 24 2007 - 17:45:26 PDT
Custom Search