fexatc requires a minimum of 4 arguments: 3 full input file
   names and 1 output file-name-prefix as command line arguments:

for example:   fexatc phenotype.txt gene_expr.txt family_id.txt outfilename

where phenotype.txt is a tab delimited (unix) text file with a one line header naming each individual,
                    and each subsequent row names a trait followed by a real number measuring
                    that trait on each individual (mathematically, the trait-by-individual matrix).

      gene_expr.txt is a tab delimited text file with a one line header naming each individual,
                    and each subsequent row names a gene followed by a real number measuring
                    that gene's expression on each individual in turn (the gene-by-individual matrix).

      family_id.txt is a tab delmited text file with a one line header, with subsequent lines
                    each giving an individual id followed by a family id (litter id for mice.)

                    To clarify the relationship to phenotype & expression files: the family_id.txt file
                    needs to have two columns, one for person id
                    and one for fam id, and that person id needs to be the header row for the
                    phenotype and expression files.

      outfilename.test.txt    will become a tab delimited text file of a matrix of test statistics
      outfilename.pvalue.txt  will become a tab delimited text file of a matrix of p-values
      outfilename.corr.txt    will become a tab delimited text file of a matrix of correlations

              each output matrix is a (# genes) x (# phenotype traits) matrix

      outfilename.sig.txt     will become a file of the significant correlations
      outfilename.allfam.txt     will become a file of the family structure
      outfilename.summary_pvalues.txt   summarizes the non-permuted  % significant pvalues at various thresholds

 optional flags which must come as the first argument if one (at most) of these is present:
    (key: t=>file is tagged with rowlabel;  c=> cancel row labelling)

          --ttt  all data files (gene expression,phenotype,family) have row-labels.
          --ttc  only the  (gene expression,phenotype) data files have row-labels.
          --tct  only the  (gene expression,phenotype) data files have row-labels (*the default*).
          --tcc  only the  (gene expression) data file has row-labels.
          --ctt  only the  (phenotype,family) data files have row-labels.
          --ctc  only the  (phenotype) data file has row-labels.
          --ccc  no files have row-labels.

          --permute=N    means permute the family (little variable) a total of N times and report
                         average p-values. All -- options must come before the file names.
          --nosave  don't save significant results, just tabulate statistics on marginal # signficant

          --numrowsexclude=N    exclude the first N rows when counting pvalues
          --numcolsexclude=N    exclude the first N rows when counting pvalues

fexatc: code Copyright(c) Jason Aten 2002-2003 <jaten@ucla.edu>. All rights reserved.
If you use this software please cite:
  Peter Kraft, Eric Schadt, Jason Aten, Steve Horvath "A family-based test for
  correlation between gene expression and trait values". American Journal of
  Human Genetics. In press 2003.

There is no expressed or implied warranty of any kind
associated with this software.