[gmx-users] Calculation of RMSF

[Justin Lemkul]

On 11/28/19 12:09 PM, Nicolas Cheron wrote:
> Dear all,
>
> I am puzzled by the way RMSF are computed. For a given trajectory (nothing
> fancy, just a regular protein):
> (1) I have computed RMSF 5 times for two frames i.e. Frame 1/2, 3/4, 5/6,
> 7/8 and 9/10, each time using the same pdb reference (if I compute RMSF for
> a single frame I sometimes have -nan, that's why I did it for couple of
> frames)

The RMSF of one frame is always undefined, because there is no fluctuation.

> (2) I have then computed RMSF for the 10 frames at the same time (1-10).
>
> I would expect that the average of RMSF^2 for (1) would equal RMSF^2 for
> (2), but that's not the case. To give you some numbers for 4 Calpha carbons:
> Avg of case (1)         case (2)
> 0,000722                  0,001673
> 0,000417                  0,001197
> 0,000400                   0,001096
> 0,000768                   0,001910
> It doesn't seem to be a rounding issue, neither a systematic difference.
>
> Can anyone please help me understand these differences?

The average structure is different, so the RMSF is different as a 
result. RMSF computes the fluctuation about an average structure, 
computed from the coordinates in the frames you analyze. So if you 
analyze 2 frames, the reference is the average of those two and you 
return the RMSF about it. If you analyze 10 frames, your reference is 
the average of those 10 and may not look anything like the average of 
any 2 frames.

-Justin

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

Justin A. Lemkul, Ph.D.
Assistant Professor
Office: 301 Fralin Hall
Lab: 303 Engel Hall

Virginia Tech Department of Biochemistry
340 West Campus Dr.
Blacksburg, VA 24061

jalemkul at vt.edu | (540) 231-3129
http://www.thelemkullab.com

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