[gmx-users] Essential dynamics analysis

Austin B. Yongye ybausty at yahoo.com
Mon Dec 7 16:58:50 CET 2009

Hi everyone,
I am very new to essential dynamics analysis, and trying to determine the similarity/dissimilarity between two different MD trajectories.

I projected the eigenvectors of trajectory-2 unto the first-two eigenvectors of trajectory-1 using g_anaeig with the -2d option and generated a plot. My very basic questions are:

1) If values on one eigenvector axis are all negative/positive, what does that mean in terms of that mode?
2) If the values are negative and positive, what does that indicate?

I then determined the inner-products (using the -inpr keyword) of the first-ten eigenvectors of both trajectories. The legend suggests that eigenvector-2 of trajectory-1 has more significant overlaps with the first-ten eigenvectors of trajectory-2. However, when I look at the projection of trajectory-2 unto eigenvectors-1 and -2 of trajectory-1, the projection unto eigenvector-2 is negative. So, I'm not sure what this means. (my question 1 above)

I also plotted the subspace overlap using the first-ten eigenvectors of trajectory-1 with all the vectors of trajectory-2, and obtained a "build-up" curve. The values go from 0-1 with increasing number of eigenvectors from trajectory-2. Does that suggest that the more modes I include from these two trajectories, the more similar their subspaces become?

How does one compute the RMSIP between two segments of a trajectory for a set of eigenvectors? Would that be done by simply determining the inner products of the eigenvectors of the both segments using the -inpr option in g_anaeig? I am not sure when the RMSIP equation comes into play.


PS. Tutorial or references will be greatly appreciated as well. I have primarily read the following papers and some book sections:
1). de Groot et al. Proteins 2009; 76;30-46
2). Amadei et al. Proteins 1999; 36;419-424
3). Amadei et al. Biopolymers 74: 448-456, 2004
4). Amadei et al. J. Comput. Chem. 18(2), 169-181, 1997
5). Roy et al. J. Mol. Graphics Modell. 27 (2009), 871-880


More information about the gromacs.org_gmx-users mailing list