[gmx-users] questions abot normal mode analysis in gromacs

[凌未风]

Dear GMX users:

I am now doing normal mode analysis of an enzyme.  I have calculated the eigenvectors by elNemo web server. Now I am  thinking of using g_nmtraj tool in gromacs to generate virtual  trajectory along a single normal mode/ several normal modes. 
I have the following questions when doing this:

To use g_nmtraj tool I may need a topology/structure file and a *.trr file as parameters for  g_nmtraj (such as    g_nmtra -s *.tpr -v eigenvec.trr -o ....). I have some problems of generating an eigenvec.trr file using my eigenvectors. 
   1, To my understanding, i need to include the initial structure as -1 frame and an average structure as 0 frame in the eigenvec.trr file, and then record eigenvectors as coordinates in PDB files and add these files to eigenvec.trr file as frames 1 to n. Do I understand the formats of eigenvec.trr correctly?
   2, The second question is what is the average structure? I only have the initial structure for NMA, what does the average structure mean here? How could I get an appropriate average structure for eigenvec.trr file and g_nmtraj analysis?
   3, As I am using elNemo web server to do NMA, I do not have the eigenvectors of the first 6 normal modes (corresponding to rotation and translation movements), so I am not including these eigenvectors to the eigenvec.trr file, I think this will not affect my analysis, right?
   4, After I generated an eigenvec.trr file by the above protocol, I get errors when I run g_nmtraj, the command I am using is 
                         g_nmtraj -s HEP-3-UofS.g96  -v eigenvec.trr -o traj.xtc -eignr 4
   And I get the following errors:  
                             Could not find mode 3 in eigenvector file.
    What does this error mean? How could I fix it?
   5, Buy using the about command, I also get an warning, which is 
                           Read non mass weighted reference structure with 5319 atoms from eigenvec.trr

      the g_nmtraj help file says:

                                   g_nmtraj generates an virtual trajectory from an eigenvector, corresponding to a harmonic Cartesian oscillation around the average structure. The eigenvectors should normally be mass-weighted, but you can use non-weighted eigenvectors to generate orthogonal motions.

        So, I guess the non mass weighted reference structure will affect my results, right? What does the non mass weighted reference structure mean, how could I get a mass weighted structure?
   6, where I can find a description of the algorithm of g_nmtraj tool?

   7, I am thinking of generating trajectory along  several normal modes simultaneously, for instance, I want to generate a  trajectory along the 7, 8, 9 normal modes together (not one trajectory  for each of these three modes). Can I do this by using g_nmtraj? I am  aware that by using "-eignr" we can define which eigenvector will be  used to generate the trajectory, but to my understanding, I can only  assign one eigenvector here, right? Or I can assign several eigenvectors  simultaneously?
   If g_nmtraj cannot do this, does any one have any idea of how to do this by other tools?

   Thanks for your time. I have too many questions. I will really be grateful if anyone can give me some advises. Look forward to hearing from you. 

Cheers, 
RXG