The case-control executable looks for haplotype motifs that distinguish
matched case and control populations.  It does this through an allele
positivity test.  P-values are calculated using Fisher's exact test and
adjusted by Bonferroni correction.

The executable is called as follows:

case-control -i <case file> <control file> [-s <motif file>]
             [-m <mutation probability>] [-p <p value>]
             [-c <min motif frequency>]
-i <case file> <control file>: file names for case and control inputs
-s <motif file>: input motif file name (default: stdin)
-m <mutation probability>: prior probability of mutation (default: 0)
-p <p value>: p-value cutoff for significance test (default: 0.001)
-c <min motif frequency>: minimum frequency of a motif and its negation
   for it to be used in association testing (default: 0)

See README.files for information on the formats of the case, control,
and motif files.

The case and control inputs take the form of pairs of haploid sequences,
each corresponding to the two haplotypes of a given individual for a
given chromosome.  For each motif, the code determines the number of
individuals whose optimal motif parse for either chromosome contains
that motif.  The code reports any motif whose counts indicate that it
is a significant predictor of case vs. control.

Output takes the following format:

[[Positive predictor start site: <start> sequence: <sequence> significance:
 <significance>] |
[Negative predictor start site: <start> sequence: <sequence> significance:
 <significance>]]*

where <start> is the index of the beginning of the motif in the set of
types sites starting from 0, <sequence> is the set of single-site
alleles corresponding to the motif, and <significance> is the
significance level of the association after correcting for multiple
hypotheses.  Positive predictor indicates a motif over-represented in
cases and negative predictor a motif over-represented in controls.