AMBER Archive (2005)

Subject: AMBER: amber file formats.

From: Shishir (shishirk_at_gmail.com)
Date: Wed Oct 19 2005 - 15:16:04 CDT


Dear all,
   My problem is this: I have a (.top+.crd) file which I want to load in xleap
to make a few changes to topology of the system. Can it be done some
other way? (since xleap doesn't loads a .top/.crd file)

I didnot know the solution for this so i cut off some code from ptraj and wrote
a (.top/.crd) to .lib converter (for ptraj reads .top/.crd files and has dumping
function for many fields), but again the .lib file format is not documented
anywhere so my code is not completely correct, in particular, the order of
.hierarchy and .connections sections in .lib file is not known to me - though
it matters little. I thought I can feed this .lib back to leap and generate a
new .lib in standard format, but leap just copies its input to output (and so
it's code - unitio.c - doesn't discuss ordering of connections/hierarchy).
probably I should look into a Pdb to .lib convertor, or somebody has a better
suggestion? On this list somebody raised the question of .lib file format
and the reply was that it is self evident. I think I'm missing something,
can anybody point out what?

Then there is the question of overall usage. Why cannot we have a single
file format throughout all of the amber? like .crd/.top files. these contain
all information for a simulation so why not have this as a standard format
which every program uses. the two reasons I see:
a) for backward compatibility : but we can provide a program that will do
  all the required conversions (and it is already to coded in some
parts of amber).
b) .top/.crd file being big: these files (including .lib files) are
bloated in their
current formats - ascii wastes quite a lot of space, a binary format
can improve efficiency of reading, manipulation, coding. for those who
would like to look
at ascii formats we can again provide the converters.

a single file format can give a lot of freedom, ease of usage. programs
can be used like a pipe then (the unix way). please comment.
(another thing that i encountered is this: the .top format of the file i had
- i donot know how it was produced - was quite a bit different from the
.top file format as listed on amber site. it may be in an older version,
but then ptraj, amber7/8 is using the same format.)

Finally I wanted to customize amber for some non standard simulation,
and so i needed to modify the code. the code this big is not so easy to
understand speedily without some help(though can be). and there is very
little documentation. Can somebody point me where i can get more
information on this. (i have done my modifications already, but still I will
appreciate better understanding of it).
suggestion for improving the documentation which will not take much
time: a) provide the dictionary for variable names and their use (this
is done at some places already). b) provide a simple pseudo algorithm
of main things (in serial/parallel version) e.g. in sander something like this:
main {
   init_calc { ... }
   loop { calc_forces; update_positions ..}
   finalize {...}
}
calc_forces { calc_energies; calc_force_from_engy_grad ...}

these two are not time taking jobs and can be done when one has
reasonable understanding of code. probably I'll do it sometime later
when i can.

thanks for your patience.
shishir.
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu