NAME

SSupdate.pl - Update the the sum and sum of squares for the bootstrap

SYNOPSIS

   SSupdate.pl [-l lrcol] [-a additive] [-d dominance] [-I hypothesis] 
         [-f SSfile] [-h] [-c] < input > output

DESCRIPTION

SSupdate.pl reads from the standard input and writes to the standard output. It can do one of two things. The first is to initialize a file of sums and sums of squares in the Ziboot.out format from a Zmapqtl.out file. The second is to read the results of a run of Zmapqtl and the current Ziboot.out file and update the results. It is a Perl script meant to be run in a loop with Prune and Zmapqtl.

OPTIONS

-f

This option requires an input filename that must exist. It allows the user to specify the Ziboot.out file for processing. If it is not given, then the script assumes that an initial file will be created.

-h

requires no operand. If used, it prints a usage message and exits.

-l

requires an integer operand. This should be the column from the Zmapqtl.out file with the likelihood ratio that you want processed. By default, it is 4.

-a

requires an integer operand. This should be the column from the Zmapqtl.out file with the additive effect that you want processed. By default, it is 0, which means that the sum and sum of squares for the additive effect will not be updated.

-d

requires an integer operand. This should be the column from the Zmapqtl.out file with the dominance effect that you want processed. By default, it is 0, which means that the sum and sum of squares for the dominance effect will not be updated.

-I

requires an integer operand. This should be an hypothesis test code. Possible values are 1, 30, 31, 32, 10 and 20. Using this option sets the proper values for the -l, -a and -d options, hence they are ignored in the presence of the -I option.

-h

requires no operand. If used, it processes the file specified with the -f option. This processing calculates the means and variances of the likelihood ratio and effects for each position. The program exits before reading from the standard input.

EXAMPLE

This Perl program was meant to be run in a shell script. Here is an example of a c shell program that allows calculate the sum and sum of squares for the likelihood ratio and additive effect in a bootstrap experiment.

  #!/bin/csh
  #   Bootstrap
  #   Copyright (C) 2000 Christopher J. Basten 
  # Usage governed by the terms of the 
  # GNU  General Public License,  version 2 or higher
  #  See the file COPYING in this directory
  #
  #   This file was meant as an example.  You  will need to edit it 
  #   to work on your particular system with your data files.
  #
  #  Start by setting the variables needed.  
  # 
  set stem=corn                       #  filename stem
  set hypo=1                          #  hypothesis for SSupdate
  set model=3                         #  analysis model
  set reps=1000                       #  number of bootstraps
  set email=basten\@statgen.ncsu.edu  #  email address for notice
  set templog=temp.log                #  temporary log file
  set qbin=/usr/local/bin             #  where are the QTL Cart binaries
  set bin=/usr/bin                    #  where are the system programs
  #
  $bin/rm -f $templog
  echo "Bootstrap experiment started " > $templog
  $bin/date >>  $templog
  $bin/echo "Stem: " $stem >> $templog
  $bin/echo "Model: " $model >> $templog
  $bin/echo "Reps: " $reps >> $templog
  $bin/echo "Email: " $email >> $templog
  $bin/mv $stem.log $stem.logsave
  $bin/rm -f $stem.z${model}a
  $qbin/SSupdate -I $hypo < $stem.z > $stem.z$model.boot
  $bin/mv $stem.z $stem.zsave
  set i=1
  while ( $i <= $reps )
  $qbin/Prune -A -V -i $stem.cro -b 1  >>&  $templog
  $bin/nice $qbin/Zmapqtl -A -V -M $model -i $stem.crb >>&  $templog
  $qbin/SSupdate -I $hypo -f $stem.z$model.boot < $stem.z > $stem.z$model.new
  $bin/mv $stem.z$model.new $stem.z$model.boot
  $bin/rm $stem.z
  @ i++
  end
  $qbin/SSupdate -I $hypo -c -f $stem.z$model.boot > $stem.z$model.booted
  $bin/mv $stem.logsave $stem.log
  $bin/mv $stem.zsave $stem.z
  $bin/echo "Bootstrap experiment ended " >> $templog
  $bin/date >>  $templog
  /usr/ucb/mail $email <  $templog

Suppose you had a data set corn.cro and a map file corn.map. To use the above shell script, create a directory called cornboot and copy the two files into it. Run Qstats on the files to initialize the qtlcart.rc file, and SRmapqtl to rank a set of markers for use with composite interval mapping. Make sure that the QTL Cartographer programs are installed in the /usr/local/bin subdirectory (or change the qbin line above). Run the bootstrap with the following command:

        % Bootstrap  &

The script will email you a message when it is complete. The example above uses interval mapping and does 1,000 bootstraps. The script above has been rewritten in Perl: Please look at the Bootstrap.pl man page for more information.

Note that the above example uses -I 1 for the SSupdate line in the loop (this is set with the set hypo=1 line in the script). This indicates that the dataset are the result of a backcross or recombinant inbred line. If you want to use a different column of likelihood ratios, you can change that option. You could create multiple files of the format Ziboot.out and collect the appropriate sums and sums of squares from different hypothesis tests by having multiple instances of SSupdate in the loop.

HYPOTHESIS TESTS

Using the -I option is an easier way to set the columns from the Zmapqtl.out file that you want to process. The following values are valid:

should be used with backcrosses or recombinant inbreds, that is only those crosses with two distinguishable marker types. It will read the likelihood ratio from column 4 and the additive effect from column 7. The dominance effect will be ignored. The likelihood ratio is for H1:H0.
can be used when more than three marker genotypes are distinguished. Likelihood ratios come from column 11, additive effects from column 7 and dominance effects are ignored. The likelihood ratio is for H1:H0.
can be used when more than three marker genotypes are distinguished. Likelihood ratios come from column 12, dominance effects from column 9 and additive effects are ignored. The likelihood ratio is for H2:H0.
can be used when more than three marker genotypes are distinguished. Likelihood ratios come from column 4, aditive effects from column 8 and dominance effects from column 10. The likelihood ratio is for H3:H0.
can be used when more than three marker genotypes are distinguished. Likelihood ratios come from column 5, aditive effects from column 8 and dominance effects from column 10. The likelihood ratio is for H3:H1.
can be used when more than three marker genotypes are distinguished. Likelihood ratios come from column 6, aditive effects from column 8 and dominance effects from column 10. The likelihood ratio is for H3:H2.

Recall that when you have two marker classes, there are two hypotheses:

H0

No QTL, that is the additive effect is zero

H1

The additive effect is nonzero.

In contrast, when you have three marker classes, then you have four hypotheses:

H0

No QTL, that is the additive and dominance effects are zero

H1

The additive effect is nonzero, but the dominance effect is zero.

H2

The dominance effect is nonzero, but the additive effect is zero.

H3

Both the additive and dominance effects are nonzero.

SEE ALSO

Zmapqtl(1), Prune(1), Qstats(1), SRmapqtl(1)

AUTHORS

In general, it is best to contact us via email (basten@statgen.ncsu.edu)

        Christopher J. Basten, B. S. Weir and Z.-B. Zeng
        Department of Statistics, North Carolina State University
        Raleigh, NC 27695-7566, USA
        Phone: (919)515-1934