Linux MolBio Software

Harry Mangalam mangalam at uci.edu
Mon Mar 18 13:57:50 EST 1996


in reference to my announcement of tacg, someone wrote to me and I thought
that I'd respond to the net as well...

>your program sounds very interesting.  However I was wondering whether
>you would know of any ftp/ww repositories of sources of other LINUX
>moloecular biology software?  Is it very difficult to port other UNIX
>molecular biology shareware to LINUX?

Yes, I've seen the queries as well and some answers (although they may
take a longer time to filter to you based on where you are in the toplogy
and connectivity of things...).

The upshot is that most tools for other unix-like OS's port to linux very
easily - it has recently beaten other commercial OS's to the punch by
becoming POSIX-compliant (see comp.os.linux.announce)

A good FTP archive to start at is Don Gilbert's ftp.bio.indiana.edu.  I'm
going to be starting my own in the next while as well, as soon as I switch
jobs.
Others are available in varying degrees at other locations; a good place
to start for Linux-specific programs is by searching the LSM entries that
are available via the net, say at: http://www.redhat.com/

  The only hiccup are those programs that were origninally writ in
FORTRAN, as there's not a very good FORTRAN compiler for LINUX.  the FSF
has recently released a sort-of-beta g77 (free f77-like compiler) and
there is a pretty good FORTRAN-to-C conversion prgram (f2c), but there's
nothing like the depth of tools for C/C++ for FORTRAN.

There is also the (sometimes slight, sometimes onerous) conversion of
pre-ANSI C code to ANSI C code, but overall, there are relatively few
problems.  ALso the availability of some required libraries (because Motif
is $ware, and is required, many Linux installations don't have it, so that
while statically linked programs will run fine, you cannot recompile the
code to optimize for your own platform unless you have bought the Motif
(or other libs).  There are also some harware dependencies - nothing
written for the Silicon graphics IRIS GL package will run on Linux, so you
can forget about many molecular modeling programs, even if the source code
was available, however, the new version of that code (OpenGL) has been
ported in software-only (so it doesn't take advantage of any hardware
acceleration - only the main cpu), so you can now begin to run some OpenGL
applications on Linux with an appropriate X server.

In my experience (mostly with Silicon Graphics, but also with DEC OSF,
SunOS, Solaris, ConvexOS), Linux is the next easiest environment to port
to other than SGI (and that's because I'm quite familiar with SGI
equipment/compilers).  For straight computation, a fast Pentium Pro with a
PCI interface and well-matched hardware should be faster than a comparable
high-end (R5000-based) Indy and the current shippoing crop of Indigo2s (w/
250 MHz R4400, although  not the newest R10000-based machines - which cost
about 3-7 times as much).

So, the upshot is: if you're looking for a platform to do biocomputing,
Linux on  fast PC hardware is definitely my 1st choice unless you have
unlimited funding or unless you absolutely require the ability to run IRIS
GL/OpenGL applications with high performance.

More questions welcome.
Cheers
Harry
-- 
Harry J Mangalam, Microbiology and Molecular Genetics, UC Irvine,
      Irvine, CA, 92717, (714) 824-4824, fax (714) 824 8598
            http://hornet.mmg.uci.edu/~hjm/hjm.html
  Computational Biology..SGI..Woodworking..Bicycling..Linux..WWW 




More information about the Bio-soft mailing list