R/qtl version 1.06-43
Karl Broman
(http://www.rqtl.org).
See http://www.rqtl.org/STATUS.txt for a list of changes. The major
changes are listed below. There is nothing particularly striking.
As always, I'd appreciate hearing any suggestions or comments or
complaints.
Note that I have moved to the University of Wisconsin-Madison.
Karl Broman
kbr...@biostat.wisc.edu
------
Major changes in R/qtl version 1.06-43:
Revised the method for calculating genotyping error LOD scores. For
each individual and each marker, the error LOD score is calculated
assuming that all other genotypes for that individual on that
chromosome are correct. The new procedure requires much more
computation time (especially in the case of dense markers), but
identifies many additional potential errors. A new argument,
version, allows one to specify use of the "new" or "old" version of
the error LOD score calculations.
Added a function geno.image for plotting an image of the genotype
data. This is much like plot.missing, but gives the genotypes in
color, rather than just black/white indicating missing/not.
Revised geno.table so that it gives reasonable p-values for the X
chromosome and for the case of dominant markers in an intercross.
Added a chr argument to obtain results for only selected
chromosomes.
Added a function cim() for performing composite interval mapping by
one of the schemes used in QTL Cartographer: forward selection at
the markers, to a fixed number of markers, followed by interval
mapping using those marker as covariates, and dropping any markers
within some fixed window around the position under test. The
results may be plotted or summarized using the functions for output
from scanone(). Also added a function add.cim.covar, for adding
dots, to a plot from plot.scanone(), to indicate the selected marker
covariates.
Extended the code for fitting the Stahl model for crossover
interference to the case of intercross data. (Modified the
functions est.map and fitstahl, and the underlying C code.)
Added functions scanoneboot and summary.scanoneboot, for deriving
bootstrap confidence intervals for the location of a QTL, but we
recommend using lodint or bayesint, instead.