AMBER Archive (2000)

Subject: Re: Help with polynucleotide building

From: David Case (case_at_scripps.edu)
Date: Wed May 24 2000 - 17:53:24 CDT


James Duncan <duncan_at_lclark.edu> wrote:

>
> I am trying to learn how to use Amber6 for the first time after spending
> a little time getting acquainted with Amber5 at the end of last summer.
> At that time I was able to build a polynucleotide from scratch by
> following the DNA tutorial on the AMBER web site (polyA-polyT 10-mer
> duplex) using nukit and nugen....
>
> Your manual states that: "Although Amber may appear dauntingly complex
> at first, it has become easier to use over the past few years, and
> overall is reasonably straitforward once you understand the basic
> architecture and option choices. Hundreds of people have learned to use
> Amber; don't be easily discouraged." I must say that I am becoming a
> bit discouraged. I have managed to get through most of the
> tutorials--including the DNA one--in the Amber6 manual but at least some
> (as mentioned above) of the additional tutorials on the Amber web site
> don't seem to work with Amber6.

  As you are finding out, Amber is (and always has been) extremely limited
in generating initial structures; it really *is* designed for those who
already have a pdb-format file for a plausible initial configuration. The
nukit/nucgen combination solves a very limited (but often useful) chore of
building a fibre-diffraction model helix for various sequences. But
generally, Amber is rarely an efficient program for problems where a
starting structure is not already available. We need to modify the "sales
pitch" at the beginning of the manual to make this point clearer.

  I believe that the "easier to use" phrase *is* more or less true for
the sorts of things Amber is really good at. Most LEaP sessions look like
the following:

     dna = loadPdb my_dna.pdb

      [optional commands like addIons, solvateBox, to add solvent]

     saveAmberParm dna prmtop prmcrd

...and you are ready to start equilibration in sander. The input to sander
and gibbs is now free-format and keyword driven, rather than the earlier
"numbers in columns" input format. These things are what I really meant
by saying "easier to use than before". Again, I should probably make the
"sales pitch" more explicit.

  The problem with web site tutorials being out of date is really our fault:
effort on documentation and coordination of examples has taken a back seat to
some latest algorithm improvement. Tom Cheatham is re-working the tutorials
relating to DNA, and we will try to clean up things on the Web site that are
left over from earlier versions....

  So how do you build DNA from scratch? Or generally solve the problem of
generating initial structures? There are a variety of alternatives:

  (a) Use a separate program designed for this, like Insight or Sybyl, etc.
      The "nucleic acid builder" program (available from my web site, see
      below) that Bill Ross mentioned is another possibility, although
      it too has its limitations, and for some people a steep learning
      curve. Other programs, like Jumna and MC-SYM, also can handle
      various nucleic acid-building tasks.

  (b) Modify LEaP to be more powerful in model-building tasks. This is
      is possible "in principle", and might be a Good Idea, but I wouldn't
      want to speculate about when/if this will ever happen.

  (c) Only work on problems where someone else (crystallographer, more
      experienced modeler...) has provided a good starting structure. This
      is mostly where my group works ... :-)

I hope these comments help....and we do apologize again for glitches in the
tutorials and documentation.

...dave case

-- 

================================================================== David A. Case | e-mail: case_at_scripps.edu Dept. of Molecular Biology, TPC15 | fax: +1-858-784-8896 The Scripps Research Institute | phone: +1-858-784-9768 10550 N. Torrey Pines Rd. | WWW: La Jolla CA 92037 USA | http://www.scripps.edu/case ==================================================================