AMBER Archive (2009)

Subject: Re: [AMBER] nab normal mode analysis has imaginary frequencies

From: case (
Date: Wed Oct 14 2009 - 13:34:05 CDT

On Wed, Oct 14, 2009, Junmei Wang wrote:

> When I ran normal mode analysis using nab for a set of ~100 residues protein
> structures, I found many of them have imaginary frequencies. Here is my
> calculation protocol:
> (1) remove waters, ions, and generate topologies
> (2) position-restrained GB minimization with a set of force constants: 20,
> 10, 5, 2, 0.0 (no restraint)
> (3) two-steps nab minimization using GB to reduce the RMS-gradient to
> ~1.0e-13
> (4) nma analysis

Can you send me (off-line) an example of one with and one without imaginary
frequencies? Do you know what fraction of the 100 proteins have this feature?

In general, what you see is not uncommon in protein work (I think the
CHARMM people have seen this a lot), but it can be hard to eliminate them
(moving along the "imaginary" eigenvector might help). There have been
some good arguments that they don't do much harm, and can probably be
ignored. But that would only be after a careful analysis, I would think.


AMBER mailing list