The latest information about Sputnik can be found at... 
http://www.abajian.com/sputnik/
---------------------------------------------------------------------------
  find repeats in fasta format seq file
  allows for indels, returns score.

  beta version.  caveat emptor.

  chrisa  29-Jul-94

  chris abajian
  University of Washington  Dept. of Molecular Biotechnology  FJ-20
  Fluke Hall, Mason Road
  Seattle WA 98195
---------------------------------------------------------------------------
Original README 
---------------------------------------------------------------------------
sputnik (beta test)

sputnik is a quick and dirty way to search for repeat
sequences with occaisional errors.  it works it's way
through the sequence looking for patterns (of lengths 2 <->
5) that repeat, allowing some degree of error by using the
familiar method of giving points for matches and
subtracting them for mismatches.  It doesn't compute an
entire identity matrix first and then pick the best of the
hits.  rather it starts at the beginning, and compares
until the score falls below a cutoff threshold.  then it
backs up to the best score that it got.  if it's above the
reporting threshold it spits out a description to stdout.

sputnik expects a file (one) in fasta format.  usage is

   sputnik <sequence file>

the test file, rep.lib, was compiled from genbank using a
string search for "Homo Sapiens" and "repeat sequences"

this is a very preliminary version.  current scoring values
are +1 for exact match, -6 for mismatch (error, insertion or
deletion).  The cutoff is -1 (i.e. falls below 0 we stop
trying) and the reporting threshold is 8.  N is counted as
a match but no point is given.  No mononucleotide repeats
are reported.

currently the maximum recursion depth is nailed to 5 for
performance reasons, although I believe that the scores
will diverge pretty rapidly anyway.  feel free to hack it
up, and let me know asap about any problems or pressing
needs.

needs:  - more complete interface, multiple files, etc.
        - search in both directions and merge overlapping
          hits ?  (Ben Koop thinks not...)
        - should allocate memory for file size, not nail
          to defined max.
        - man page
        - handle other ambiguities besides 'N' ?

-chris