- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Jason Swails <jason.swails.gmail.com>

Date: Wed, 29 Aug 2012 10:38:46 -0400

On Wed, Aug 29, 2012 at 1:49 AM, Catein Catherine <askamber23.hotmail.com>wrote:

*>
*

*> Dear Sir/Madam,
*

*> I was trying to calculate the RMSF for a trajectory. Since it is a long
*

*> trajectory (1000 ns), I cannot do the cpptraj analysis in one-goes with
*

*> supercomputer with only 24G RAM. I have to do it step by step. However,
*

*> I have a problem, if I want to plot RMSF (per residues) vs residue number
*

*> for the whole trajectory, for instance if I got three profiles 1-300 ns,
*

*> 301 to 600 ns and 601 to 1000 ns. What should I do to combine all the
*

*> three profile into one single profile that represent the whole trajectory?
*

*>
*

Look at the RMS formula ( sqrt( sum(x^2) / N) ) and try to combine each

section such that the formula you use is ultimately equivalent to the

formula for the RMSF of the entire trajectory.

So for instance, for each data point, square the RMSF value, multiply by

the number of frames in that window (this value should be equal to sum(x^2)

now), then add up for all 3 windows. Then divide by the total number N

over all 1000ns and take the square root to arrive back at the RMSF formula

for the whole trajectory.

You could also try extracting every 5th snapshot from the original

trajectory and performing the whole analysis in one run.

HTH,

Jason

Date: Wed, 29 Aug 2012 10:38:46 -0400

On Wed, Aug 29, 2012 at 1:49 AM, Catein Catherine <askamber23.hotmail.com>wrote:

Look at the RMS formula ( sqrt( sum(x^2) / N) ) and try to combine each

section such that the formula you use is ultimately equivalent to the

formula for the RMSF of the entire trajectory.

So for instance, for each data point, square the RMSF value, multiply by

the number of frames in that window (this value should be equal to sum(x^2)

now), then add up for all 3 windows. Then divide by the total number N

over all 1000ns and take the square root to arrive back at the RMSF formula

for the whole trajectory.

You could also try extracting every 5th snapshot from the original

trajectory and performing the whole analysis in one run.

HTH,

Jason

-- Jason M. Swails Quantum Theory Project, University of Florida Ph.D. Candidate 352-392-4032 _______________________________________________ AMBER mailing list AMBER.ambermd.org http://lists.ambermd.org/mailman/listinfo/amberReceived on Wed Aug 29 2012 - 08:00:04 PDT

Custom Search