Bootstrapping

The -b option with a value of 1 tells Prune to create a single bootstrapped data set. This option should be used alone. It will sample the data set with replacement, creating a new data set of the same sample size and writing it to the file qtlcart.crb. Of course, you can change the output file name by changing the output filename stem with the -o option. Using Prune, one can perform a bootstrap experiment on the data set. This is much easier to do on a UNIX workstation than a Macintosh or MS-Windows machine because it can be automated in a batch file.

Suppose qtlcart is the filename stem, and use $m$ as the integer code for the model, $h$ for the hypothesis test, and $r$ for the number of repetitions, then this snippet of psuedo-code will do a bootstrap analysis

Zmapqtl -A -V -X qtlcart -M $m$
SSupdate.pl -I $h$ $<$ qtlcart.z $>$ qtlcart.z.boot
move qtlcart.z qtlcart.z.save
$i = 1$
while $( i < r )$ {
Prune -A -V -i qtlcart.cro -b 1
Zmapqtl -A -V -M $m$ -i qtlcart.crb
SSupdate.pl -I $h$ -f qtlcart.z.boot $<$ qtlcart.z $>$ qtlcart.z.new
move qtlcart.z.new qtlcart.z.boot
delete qtlcart.z
$i = i + 1$
}
SSupdate.pl -I $h$ -c -f qtlcart.z.boot $>$ qtlcart.z.booted
move qtlcart.z.save qtlcart.z

Except for the while loop line, he symbols $<, >, »$ above are for input and output redirection. SSupdate.pl is a Perl script available with the distribution of the programs (look in doc/scripts subdirectory). Note that the work is done in the while loop. For each repetition, a bootstrapped data set is created with Prune. This data will be placed in the file ending with .crb. Zmapqtl then analyzes this bootstrapped data. The script SSupdate.pl reads the results of the Zmapqtl run and updates a file with the sum and sum of squares of the test statistic and estimates of effects. Also, SSupdate.pl is run to initialize the boot file, so you will need to run Zmapqtl on the original data before doing the bootstrap. When this is finished, the script runs SSupdate.pl with the -c option to get the mean and variance of the likelihood ratio, additive effect and dominance effect at each test site. There is a C shell script called Bootstrap in the scripts subdirectory of the distribution that implements the above idea.