Sorry I didn't see this earlier! Here's what happening. The energies are
not being extracted in random order, but rather an order which matches the
resulting trajectory. In this sense, the energies will be properly
attributed to each conformation so long as you read the energies_mdgx.dat
file along with the trajectory. Try converting the coordinates Coords.cdf
to PDB / ascii and verifying that the frames match:
cat > conv.ptrj << EOF
trajin Coords.cdf
trajout Coords.pdb pdb
go
EOF
There should be no problem, I've checked over this specifically. Back when
I did this same chore with a python script, the ordering got mangled by the
imported glob library. When I actually sat down to write it in proper C
code, I began to see the issues. At bottom, the computer doesn't
alphabetize or numeratize things the way you do. If you wish to see under
the hood, look in AmberTools/src/mdgx/Parse.c for DirectoryFileSearch(),
and find the following:
while ((ent = readdir(dir)) != NULL) {
...
}
It's that ent = readdir() operation that I'm reliant upon, which glob is
probably reliant upon, and which you might use yourself if you need to
write a directory scanning function at some time in the future. It will
list each file in the directory, but not in the order you might expect. I
simply make sure to extract coordinates and energies from one file at a
time and keep them in the order they were read, while culling the bad data
points.
Try running a simple parameterization with energies_mdgx.dat and
Coords.cdf, and you should be able to get sane results (errors <= 3-4
kcal/mol once you include all the parameters relevant to the configurations
you've generated). Let me know if you don't!
Dave
On Mon, Jun 12, 2017 at 9:36 AM, Charles-Alexandre Mattelaer <
camattelaer01.gmail.com> wrote:
> Dear Amber users
>
> As I already mentioned, I'm using mdgx to generate parameters for modified
> nucleic acids. However, after extraction of the energies calculated by
> ORCA, the resulting 'energies_mdgx.dat' file listed the calculated energies
> in random(?) order.I'm not quite sure what is happening here since the
> extraction should be fairly simple (enclosed file '04_ExtractEnergies').
> Using a 'for'-loop with the command sed I generated my own
> 'energies_sed.dat' file for comparison.
>
> I enclosed the first of the ORCA outputs (Conf1.oout) so you can verify
> that this energy does not show up as the first entry in the
> 'energies_mdgx.dat' file. I also enclosed my own extracted
> 'energies_sed.dat' file where this is correctly performed.
>
> Additionally, I can add that during the generation of the conformers (also
> using mdgx) I lose several conformations so that I don't have 'Conf1.orca'
> to 'ConfN.orca' consecutively, resulting in several empty 'ConfX.oout'
> files since ORCA is called consecutively in a 'for'-loop. I don't know if
> this should bother mdgx during extraction to throw the energies around?
>
> I assume this influences the parameter fitting stage, since the energies
> should be attributed respectively to the conformation in the trajectory
> file?
>
> Kind regards
>
> Charles-Alexandre Mattelaer
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Jun 12 2017 - 21:30:02 PDT
