AMBER Archive (2006)

Subject: Re: AMBER: Question about NSCM

From: David A. Case (case_at_scripps.edu)
Date: Tue Jul 11 2006 - 10:02:11 CDT

On Tue, Jul 11, 2006, Xiaowei (David) Li wrote:

> I am currently trying to calculate the vibrational modes as well as
> the absorption spectra in THz frequency range (10-30 cm^-1) of DNA
> decamer. My procedure of simulation starts performing 200ps molecular
> dynamics of DNA decamer in explicit water with NVE ensemble after
> equilibration as described in the AMBER tutorial website. Then the
> normal modes of vibration are calcualted using the quasi-harmonic
> analysis which is implemented in PTRAJ program. I have found that the
> vibrational modes are quite sensitive to the choice of NSCM,
>
> NSCM=1: 10.05 10.86 11.39 12.19 12.64 13.17 14.3
> 14.7 15.21 15.97 16.14 16.88 17.27 18.26 18.55
> 19.48 20.06 20.55 20.86 21.87 22.43 23.33 23.53
> 24.15 24.31 25.35 25.95 26.41 27.05 27.83 28.09
> 28.3 29.18 29.94 30.25;
> NSCM=100: 10.56 11.07 11.32 11.87 12.83 12.93 13.3
> 13.81 14.73 15.31 15.45 15.81 17.16 17.86 18.25
> 18.84 19.19 19.86 20.65 20.88 21.81 22.06 22.98
> 23.48 24.16 24.72 25.02 25.95 26.69 26.74 27.3
> 27.73 28.93 29.25 29.85 30.25 30.44 30.77;
> NSCM=1000: 10.15 10.95 11.16 12.13 13.06 13.51
> 14.28 14.57 15.27 15.29 15.58 16.15 17.02 17.95
> 18.58 19.08 19.65 19.86 21.08 21.7 22.08 22.55
> 23.1 23.52 23.93 24.88 25.28 25.9 26.05 26.74
> 27.65 27.84 28.6 29.16 29.63 30.24 30.59

Are you sure that this is not just variation among runs? I'm actually
surprised that your frequencies are so close to each other for the three runs
after only 200 ps of sampling. Your frequencies are only different by
fractions of a cm**-1. Getting true convergence will certainly require much
longer sampling. I don't see why the value of NSCM should be important to
this calculation, and I don't see any direct evidence from your results that
it actually *is* important.

...dac

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu