[gmx-users] radius of gyration

[Justin A. Lemkul]

On 5/8/12 1:06 PM, dina dusti wrote:
> Dear GROMACS Specialists,
>
> I have one question about radius of gyration for micelle, Please help me.
> My micelle is created at 100000 ps by Martini CG force field.

Does this mean that during the first 100 ns, there is no micelle?  That is, it 
is only completely formed after 100 ns have passed?

> When I calculate the radius of gyration from this time as "g_gyrate -f 1.xtc -s
> 1.tpr -n 1.ndx -o gyrate.xvg -b 100000" and "g_analyze -f gyrate.xvg -av -ee",
> my answer is:
>
> std. dev. relative deviation of
> standard --------- cumulants from those of
> set average deviation sqrt(n-1) a Gaussian distribition
> cum. 3 cum. 4
> SS1 1.811581e+00 1.586248e-02 3.708049e-05 0.399 0.318
> SS2 1.476027e+00 5.355297e-02 1.251867e-04 0.009 -0.039
> SS3 1.480446e+00 5.226397e-02 1.221735e-04 0.004 -0.060
> SS4 1.478287e+00 5.332558e-02 1.246551e-04 -0.005 -0.084
>
> Set 1: err.est. 0.000281889 a 0.468124 tau1 20.1606 tau2 308.225
> Set 2: err.est. 0.0024333 a 0.925872 tau1 536.208 tau2 8592.94
> Set 3: err.est. 0.00190226 a 0.971916 tau1 544.756 tau2 7044.58
> Set 4: err.est. 0.00267771 a 0.991392 tau1 629.613 tau2 88302
>
> But when I calculate it from start time of simulation as "g_gyrate -f 1.xtc -s
> 1.tpr -n 1.ndx -o gyrate.xvg" and "g_analyze -f gyrate.xvg -av -ee", my answer is:
>
> std. dev. relative deviation of
> standard --------- cumulants from those of
> set average deviation sqrt(n-1) a Gaussian distribition
> cum. 3 cum. 4
> SS1 1.979407e+00 6.228274e-01 1.392685e-03 2.317 4.169
> SS2 1.613785e+00 5.401095e-01 1.207722e-03 2.478 4.949
> SS3 1.601265e+00 4.713569e-01 1.053986e-03 2.547 5.460
> SS4 1.624623e+00 5.387790e-01 1.204746e-03 2.208 3.719
>
> Warning: tau2 is longer than the length of the data (1.2e+06)
> the statistics might be bad
> invalid fit: e.e. -nan a 5.56656 tau1 674505 tau2 8.75077e+07
> Will fix tau2 at the total time: 1.2e+06
> a fitted parameter is negative
> invalid fit: e.e. -nan a 18.789 tau1 806548 tau2 1.2e+06
> Will use a single exponential fit for set 1
> Set 1: err.est. 0.212562 a 1 tau1 69885.8 tau2 0
> a fitted parameter is negative
> invalid fit: e.e. 20.1017 a 2.22647 tau1 210523 tau2 -6.77257e+08
> Will fix tau2 at the total time: 1.2e+06
> a fitted parameter is negative
> invalid fit: e.e. -nan a 3.04296 tau1 249121 tau2 1.2e+06
> Will use a single exponential fit for set 2
> Set 2: err.est. 0.170009 a 1 tau1 59447.7 tau2 0
> a fitted parameter is negative
> invalid fit: e.e. 0.139564 a 1.61177 tau1 36947.7 tau2 11359.6
> Will fix tau2 at the total time: 1.2e+06
> a fitted parameter is negative
> invalid fit: e.e. -nan a 2.35981 tau1 195092 tau2 1.2e+06
> Will use a single exponential fit for set 3
> Set 3: err.est. 0.149959 a 1 tau1 60729.6 tau2 0
> a fitted parameter is negative
> invalid fit: e.e. 6.4854 a 5.92085 tau1 759910 tau2 -1.67528e+07
> Will fix tau2 at the total time: 1.2e+06
> a fitted parameter is negative
> invalid fit: e.e. -nan a 30.0389 tau1 922726 tau2 1.2e+06
> Will use a single exponential fit for set 4
> Set 4: err.est. 0.184771 a 1 tau1 70566.6 tau2 0
>
> and my plot of errest.xvg in the first one is similar with article "B. Hess, J.
> Chem. Phys. 116:209-217, 2002" but in the second one is different whereas
> err.est in second are similar to micellar articles.
>
> I don't know which one is true, time of simulation start or time of creation of
> micelle?
>
> Both of them (the radius of gyration) is in agreement with experimental but
> second one is more near than first one.
>

If I am interpreting your situation correctly (and please do answer the question 
posed above), then the second case (which leads to an unreliable error estimate) 
incorporates 100 ns of time in which no micelle exists, so you're measuring 
something that you shouldn't, as it includes a substantial amount of time in 
which diffuse molecules will artificially increase the value of Rg if they are 
spread throughout the solvent.  If after 100 ns, a micelle exists, then the 
first command is appropriate to measure its radius of gyration.

-Justin

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================