Links to project.eigen were added in related
fucntions docs.
doMinimize was added to compute.pigengene.
More diagnostic messages from scoreCandidates.
Using "/" is avoided in determine.modules and
learn.bn functions to support running them on Windows.
The "toLocal" task in learn.bn() and sbatch()
still do not support Windows.
inst/script/bn.calculation.job.R is added as an
example for the bnCalculationJob argument.
Setting toCompact to TRUE in
make.decision.tree() has the same effect as NULL.
The averaged.network() function now correctly
passed test data to the module.heatmap() function.
More details on the effect of test data in the documentation of
one.step.pigengene() and make.decision.tree()
functions. make.decision.tree() returns more information on
performance.
Scaling is done in the compute.pigengene() function
to avoid the following error that Sogand detailes on her lano:
Error in La.svd(x, nu, nv) : error code 1 from Lapack routine 'dgesdd'
The compute.pigengene() function does not warn about
almost constant genes.
Constant genes are now ignored in the
compute.pigengene(doWgcna=FALSE,...) function to prevent a
run time error.
The default value changed in
gene.mapping(leaveNA=FALSE, ...).
The doWgcna option is added to the
compute.pigengene function.
The get.enriched.pw function
adds gene symbols in the excel file.
The beta table can now be suppressed in the
determine.modules function.
Habil changed the default value from
hu.mouse(host="useast.ensembl.org", ...) to
hu.mouse(host="www.ensembl.org", ...) to
prevent a check error on Bioconductor.
Habil added the doReturNetworks argument to one.step.pigengene().
Habil renamed identify.modules() to
determine.modules().
Neda exported identify.modules(),
make.filter(), and apply.filter() functions.
Neda added the identify.modules(),
make.filter(), and apply.filter() functions, but not
exported them yet.
The averaged.network() function does not
have the nodes argument in bnlearn Version >=4.7, and
thus this argument was removed.
The get.enriched.pw() function is added.
A bug fix in the gene.mapping() function that used to occur
when we had multiple output databases.
The message.if() function can now write the message
in a text file.
The C50 plot function seems to have different behavior when the number of Labels is 2. Habil reverse the color to fix the resulting bug.
When building the Ubuntu, Neda got an error message on building the vignette "argument is of length zero". It was due to her old version of BiocStyle, 2.14.4. To solve this issue, Pigengene now depends BiocStyle Version >= 2.19.1. Even 2.18.1 might be enough with R 4.0.3.
When automatically computing the threshold in the
consensus function, we now do not allow it to be more than
1 to be compatible with bnlearn 4.6.1.
The modules can now be determined in the
module.heatmap function using the new mes argument.
combine.networks now really removes the big TOM file
when doRemoveTOM=TRUE.
The DiseaseColors argument in the
plot.pigengene function can now be set automatically.
pheatmap.type now works fine even when the number of
samples in a condition is only 1.
one.step.pigengene now talks more, and automatically
converts input data frames to matrices.
module.heatmap now talks more, and does not get stuck
if a module has only one gene, or if the expression levels of some
genes in a module are not available in the data.
inherits(a, b) is now used instead of class(a)==b.
doSave was added to the combine.networks() function.
In the combine.networks() function, the netwok
typo was fixed.
The pipeline is now explained step by step in the vignette.
The Data argument of compute.pigengene can now
be a matrix with only 1 column.
The doRetuNetworks argument is now added to the
combine.networks function.
message.if() is now exported.
Better QC in the gene.mapping() function, the
possible keys will be printed if the input is not appropriate.
save.if() is now exported.
In the combine.networks() function,
selectedModules does not need to be "All". Also, if
saveFile=NULL, nothing will be saved without any error.
Data and Labels can now be lists, which will be combined using
combine.networks() before analysis.
repeat.data(times=1,...) now produces valid output.
In the compute.pigengene function, if a module has
only one gene, its name is now not omitted in the csv file.
An object of pigengene-class can now have a
heavyToLow attribute.
In the compute.pigengene function, the columns of
Data and names of modules can now differ.
The dOrderByW argument is now added for the compute.pigengene function.
All of the learn.bn function can now be set from
one.step.pigengene through bnArgs.
gene.mapping can now output multiple conventions.
The version of the package C50 is now required to be
at least 0.1.2, which exports the as.party.C5.0() function.
stats::cor is used in compute.pigengene and
draw.cor functions. See the NAMESPACE for the important reason.
cor is imported from WGCNA, but not from stats, because WGCNA does
not call the cor function properly.
RsquaredCut is added to the arguments of the one.step.pigengene function.
In the get.fitted.leaf function, the function
C50:::as.party.C5.0 is used, which used to be exported in the previous
versions of C50, but not in version 0.1.1.
The check.nas function is now exported.
Order of conditions in pheatmap.type can now be determined by the user.
A bug in gene.mapping () function fixed to better map probe IDs.
The doTranspose argument added to the heatmap.type() function.
A bug in the module.heatmap() function fixed.
The mouseHomologFilter argument is added to the unexported furntion hu.mouse() with a default value of "with_mmusculus_homolog". This is now compatible with biomaRt_2.30.0, it used to be "with_homolog_mmus" in older versions of biomaRt.
Checking the pigengene input of module.heatmap().
Issues in the balance() function (not exported) where Labels is a factor are resolved. Also, if all sampls have the same size, oversampling is automatically turned off.
If Labels is a factor, it is now converted to a character vector in check. pigengene.input().
The module.heatmap() function now has the doAddEigengene and scalePngs arguments.
The compute.pigengene() function now reports also the size of modules in the pigengene_pvalue.csv output file.
The compute.pigengene() function now uses welch.pvalue() instead of pvalues.manov().
Under review by Bioconductor.
Created.