MultiRegress uses stepwise regression to map quantitative trait loci. The data consist
of trait values that will be mapped onto expected genotypes. The standard data set can
be translated by JZmapqtl using model 9. Map information is encoded in the
data file and thus a separate map is not needed.
You might rightly ask ``What does this program add to the QTL Cartographer system?''
First, it doesn't require a map like the
other programs in the QTL Cartographer system. Second, since all of the
genotypic expected values have been calculated, just about any type of cross
could be analyzed. The user could write a program to calculate
expected genotypes at specified sites. Finally, it can speed up the process of
finding QTL when using MImapqtl (see the EXAMPLE section).
See QTLcart(1) for more information on the global options
-h for help, -A for automatic, -V for non-Verbose
-W path for a working directory, -R file to specify a resource
file, -e to specify the log file, -s to specify a seed for the
random number generator and -X stem to specify a filename stem.
The options below are specific to this program.
If you use this program without specifying any options, then you will
get into a menu that allows you to set them interactively.
Use this to specify which trait MultiRegress
will analyze. If this number is greater than
the number of traits, then all traits whose names do not begin with a minus sign
will be analyzed. If 0, then no traits except those beginning with a plus sign
will be analyzed. The default is
to analyze trait 1 only.
# 1002909319 -filetype JZmapqtl.zr
# QTL Cartographer v. 1.15e, October 2001
# This output file (qtlcart.zr) was
# created by JZmapqtl...
# It is 13:55:19 on Friday, 12 October 2001
# This output of JZmapqtl is meant to be used
# with MultiRegress
-walk 2.00 Interval distance in cM
-cross B1 Cross
-otraits 1 Number of explanatory variables
-traits 2 Number of Traits
-positions 39 Number of positions
-n 9 Sample Size
-Trait 1 Trait.1
5 5.3 6.2
5.8 6.7 6.1
-Trait 2 Trait.2
15 15.3 16.2
25.8 16.7 26.1
-Otrait 1 Sex
1 2 1 1 1 2 2
-Site 1 -parameter additive -chromosome 1
-marker 1 -name c1m1 -position 0.000100 -values
0.5 0.5 0.5
-0.5 -0.5 -0.5
-Site 2 -parameter additive -chromosome 1
-marker 1 -name c1m1 -position 0.020100 -values
0.4984 0.4984 0.4984
-0.4984 -0.4984 -0.4984
-Site 3 ......
The data file above was created by JZmapqtl with model 9. The
header of the file is similar to the qtlcart.cro format: The first
line has a long integer and specifies the filetype as JZmapqtl.zr.
Some header information is followed by parameter definitions that
include the distances between sites (same as the walking speed in
Zmapqtl, JZmapqtl and MImapqtl), the cross, numbers of
categorical traits and quantitative traits, positions (or sites) and
sample size (n). The data set above has a sample size of 9 for two
traits, one categorical trait and 39 genotype sites. The cross and
walk parameters are not needed by MultiRegress: They are provided
as a reminder of how the data set was created. The genotypes are
expected QTL types based on flanking marker information.
After the parameters, the traits are listed. For each trait, there will
be a token -Trait followed by the trait number, trait name and n
real values. After the traits come the categorical traits in the same
format: The token -Otrait is followed by the categorical trait
number, name and then n integer values.
Finally, data for each of the sites are presented. Site data start
with the token -Site followed by information about the site. The
token -parameter is followed by the word additive or dominance
indicating what expected value is calculated. The other tokens indicate
which chromosome and left-flanking marker define the site, and the
position is from the left telomere of the chromosome. The token
-values is followed by n expected values of the QTL genotype at
the site. This structure is repeated for each site.
Suppose we have a data set for an SF3 population in qtlcart.zr with three traits and the filename
stem has already been set to qtlcart.
% MultiRegress -I 30 -t 4
Does a stepwise regression with backward elimination steps
for the dataset. All three traits are analyzed and both additive and dominance
effects are estimated.
One can also speed up the process of finding QTL using multiple interval mapping.
The core algorithms of MImapqtl are very compute intensive. As an example,
using MImapqt to search for QTL de novo takes 934 seconds on a Macintosh G4
with an 867 MHz processor. Contrast this with the following sequence:
Converting the data with JZmapqtl and searching for putative QTL with MultiRegress
yields a starting point for MImapqtl. Rqtl translates the output of
MultiRegress so that MImapqtl can use it as an initial model. The -p 1 option tells
MImapqtl to set the phase variable to one, and thus the program expects the input
model to be in mletestPhase0.mqt. This method takes about 25 seconds and comes up
with a very similar set of QTL as using MImapqtl to search from scratch.
If you have a multitrait data set, then use all of the traits. Convert them all with
JZmapqtl by using a trait value greater than the number of traits, and be sure that
none of the traits have names beginning with a minus sign.
Christopher J. Basten, B. S. Weir and Z.-B. Zeng
Bioinformatics Research Center, North Carolina State University
1523 Partners II Building/840 Main Campus Drive
Raleigh, NC 27695-7566 USA