| news | R Documentation |
Version 2.0-x of wgaim is here and it celebrates
a updated release of the package that utilizes the linear mixed
modelling functionality of the R package ASReml-R V4. It
should be noted that this version of wgaim is not compatible
with ASReml-R V3 and users should revert to version 1.4-11 if
a compatible version is required. Within this new version of
wgaim there have been many subtle changes to functions and
their arguments. Most of these changes have been documented below.
The code in the main calling function wgaim.asreml() has been
significantly streamlined for better integration with new features
of ASReml-R V4. Many adjunct wgaim functions such as
mergeData() and updateWgaim() have been removed. Model
updating now occurs directly using update.asreml().
To ensure consistency between the phenotypic data used in
fitting the base model and the phenotypic data used in wgaim.asreml(),
the phenoData argument has been removed from the
wgaim.asreml call. The data is now recalled through
backwards evaluation of the base model call.
The wgaim can now handle "f2" cross objects. This
includes the appropriate imputation of missing allele values through the
functionality of cross2int().
Some of the cross2int() functions arguments have been
changed to more appropriately reflect the nature of the task being
implemented. Specifically, argument missgeno has been changed to
impute and rem.mark has been changed to
consensus.mark.
All link.map.xxx functions have changed to
linkMap.xxx for better naming consistency with S3 methods.
The out.stat function has changed to outStat and has
been completely rewritten to use ggplot2 functionality. See
?outStat for complete details.
A vignette title "An Quick Introudction to wgaim QTL Analysis" is available with the package and can be viewed using:
vignette("wgaim_intro")
Fixed a bug that caused summary.wgaim() when only one
QTL was found.
Removed maxiter = 1 from internal
predict.asreml() to prevent spurious output of
non-convergence warnings.
The use of "." in the chromosome names muddles regular
expression string matching in parts of the wgaim call. There is now
code in cross2int() to remove "."s from any chromosome names.
Flanking marker highlighting in linkMap.wgaim() was
incorrect and has been amended.
Restrictions on the layout of the plots in out.stat()
have been removed.
The argument flanking has been added to the QTL
plotting functions ro ensure that only flanking markers or
linked markers are plotted and highlighted on the linkage map.
The forward selection algorithm has been accelerated further by smart matrix decomposition of the relationship matrix. Users can expect around a 35% reduction in computation time.
Outlier statistics and BLUPs can now be returned for any
iteration of the algorithm regardless of whether a
significant QTL is detected. This now allows easy access to
outlier statistics and BLUPs for the first iteration when no QTL
are detected (see the breakout argument of wgaim.asreml().
The package now includes a PDF reference manual that is accessible by
navigating to the "doc" directory of the package. This can
be found on any operating system using the command
> system.file("doc", package = "wgaim")
The reference manual contains WGAIM theory and two thorough examples that show the features of the package. It also contains a "casual walk through" the package providing the user with a series of 5 steps to a successful wgaim analysis.
The package now includes three fully documented phenotypic and genotypic data sets for users to explore. Two of these three have been used in the manual and scripts that follow the examples in the manual are available under the "doc" directory of the package.
The package now provides very efficient whole genome QTL analysis of high
dimensional genetic marker data. All genetic marker data is passed
into wgaim.asreml() through the "intervalObj"
argument. Merging of genotypic and phenotypic data occurs within
wgaim.asreml().
wgaim.asreml() has several new arguments related
to selection of QTL. The "gen.type" argument allows the user to
choose a whole genome marker analysis or whole genome mid-point
interval analysis from Verbyla et. al (2007). The "method" argument gives you the choice of placing
QTL in the fixed part of the linear mixed model as in Verbyla et.al
(2007) or the random part of model as in Verbyla et. al
(2012). Finally, the "selection" argument allows you to choose whether QTL selection
is based on whole genome interval outlier statistics or a two stage process of
using chromosome outlier statistics and then interval outlier
statistics.
A "breakout" argument is now also provided which allows
the user to breakout of the forward selection algorithm at any
stage. The current model along with any calculated QTL components are
all available for inspection.
All linkage map plotting functions can be subsetted by predefined distances. This includes a list of distances as long as the number of linkage groups plotted.
Fixed a bug that created 0s for NAs in linkage groups with one marker.
Fixed a bug that caused wgaim.asreml() to bomb out if
the number of markers was less than the number of genotypes.
Fixed a bug that outputted warning messages regarding a
NaN calculation from sqrt(vatilde) in qtl.pick().
Fixed a bug that caused wgaim to crash if method =
"random" was used with the new version of asreml.
Fixed a bug that caused wgaim to crash with very recent versions of asreml (04/05/2015).
cross2int() now accepts R/qtl objects with cross type
"bc","dh","riself".
Fixed an issue with the internal function fix.map() that
allowed some co-located sets of markers to appear in the final
reduced linkage map.
Fixed a long standing scoping issue with different versions of ASReml-R.
Fixed an elusive problem that causes wgaim models to increase the size
of your .RData upon saving. This is actually an inherent problem with
using model formula within a function a returning the subsequent
model. There is now a function at the very tail of
wgaim.asreml() that quietly destroys the useless environments
that these formula contain.
Fixed bug that caused wgaim.asreml() to crash when no QTL
were found.
Fixed bug that caused summary.wgaim() to crash when one
QTL was found using method = "random".