exportBins(..., format = "vcf") did not add meta-data information
for field END to the VCF file header.poolRuns() would give an error "Error in colMeans2(oldphenodata,
cols = numericCols, useNames = FALSE) : Argument 'x' must be a
matrix or a vector."Fix markup typo in help("exportBins").
Add package test for poolRuns().
class(x) == "data.frame" rather than
inherits(x, "data.frame").segmentBins() would report the sample names as "NA" in output
messages if the sample name contained hyphens, or other symbols
automatically replaced by data.frame(..., check.names = TRUE).
This was a harmless bug.multisession
future strategy, instead of the deprecated multiprocess strategy.seeds of segmentBins() is defunct. It has been
deprecated and ignored since QDNAseq 1.21.3 (September 2019).Now argument logTransform of exportBins() is ignored if type = "calls".
Now exportBins() returns the pathname to the files written.
Test code coverage was increased from 42% to 52%.
Add package test for exportBins().
exportBins(fit, format = "seg", ...) and format = "vcf" would
merge segments with equal copy-number calls if they were
interweaved with copy-neutral segments.
exportBins(fit, format = "seg", ...) and format = "vcf"
produced an obscure error with messages Error in dimnames(x) <- dn : length of dimnames [2] not equal to array extent for samples
with no copy-number abberations.
exportBins(fit, format = "seg", file = ...) and format = "vcf"
did not respect argument file but instead wrote files of its own
names to the current working directory.
exportBins() would corrupt option scipen. Now it is left
unchanged.
callBins() produces warnings on Recycling array of length 1 in vector- array arithmetic is deprecated. Use c() or as.vector() instead. in R (>= 3.4.0). This is a problem in the package
CGHcall dependency and is something that needs to be fixed
there. For further details, please see
https://github.com/tgac-vumc/CGHcall/issues/2.Test code coverage was increased from 32% to 39%.
Added package tests for binReadCounts().
binReadCounts() would fail when specifying argument chunkSize.
The fix was to require future package version 1.22.1 or newer.binReadCounts().plot(..., logTransform=TRUE, doSegments=TRUE) on QDNAseqSignals would position segments that were out of range incorrectly, because it forgot to take the log transform on those outliers.verbose. For backward compatibility, it currently defaults to
'verbose=TRUE' but that may be changed to 'verbose=FALSE' in a
future release.callBins() now respects option QDNAseq::verbose for controlling
whether output from the CGHcall package should be relayed or
not.
MEMORY: Utilize more memory-efficient matrixStats functions
colSums2(), colMeans2(), etc.
seeds of segmentBins() is deprecated because it did
not use proper parallel random number generation (RNG). We now
instead rely on future.apply::future_lapply(..., future.seed=TRUE) for this.Package now imports the future and future.apply packages; previously the future was listed as a suggested package.
Package no longer depends on BiocParallel.
binReadCounts() now uses the future framework instead of
BiocParallel for parallelization.
Using future_lapply() and future_apply() of the well-tested
future.apply package instead of internal analogue
implementations.
TESTS: Now testing numerical reproducibility also for parallel
processing (using future strategies multisession and
multicore).
TESTS: Now asserting numerical reproducibility of also
segmentBins() and callBins().
exportVCF() is no longer exported. Use exportBins(..., format="vcf") instead.VCF and SEG file export have been implemented to allow use of downstream analysis tools such as Cartegenia (NGS) Bench.
binReadCounts() now supports parallel computing.
calculateBlackListByRegions() has been implemented for convient
bin overlap calculation of any set of regions.
estimateCorrection(), segmentBins(), createBins(), and
calculateBlacklist() now support parallel computing (see vignette
for more details)
callBins() can now also use cutoffs instead of CGHcall.
binReadCounts() now contains parameter pairedEnds to specify
when using paired-end data, so that expected variance can be
calculated correctly
segmentBins() now allows seeds for random number generation to be
specified for reproducibility.
binReadCounts() supports chunked processing of BAM files.
estimateCorrection() now also allows correcting for only GC
content or mappability.
applyFilters() and highlightFilters() now work properly when
using a numerical value for parameter residual.
highlightFilters() no longer highlights entire chromosomes for
which the residual filter is missing altogether, which matches the
behavior of applyFilters().
getBinAnnotations() now allows custom bin annotations to be
loaded via the path parameter even when an annotation package has
been installed.
phenodata files with a single variable are now handled correctly.
calculateMappability() now retains correct chromosome order even
when bigWigAverageOverBed reorders them.
calculateBlacklist() now correctly handles non-integer chromosome
names.
chromosomes() no longer tries to return chromosome names as an
integer vector, but returns a character vector instead.plot() and calculateMappability() now work also for other
chromosome names besides numbers and "X", "Y", and "MT".createBins() now properly selects chromosomes when ignoring
mitochondrial DNA. Please note that the mitochondrial DNA is only
ignored when it is called either "chrMT", "chrM", "MT", or
"M".correctBins() now filters out bins with missing loess correction
estimate.createBins() now also works for species not supported by
GenomeInfoDb (such as Canis lupus familiaris).applyFilters() now ignores the residual filter for chromosomes
for which it does not exist (instead of always filtering out the
entire chromosome).Add an asterisk in the noise measure to clarify it's not a regular standard deviation or variance, but first scaled with the mean (so that the mean is 1.0 and the relationship holds between variance and 1/N, where N is the average number of reads per bins).
Clarify the use of log2/sqrt transformation in segmentBins().
segmentBins() supports another transformation option besides
log2: sqrt(x + 3/8), which stabilizes the variance.
plot() can skip plotting of segments and calls by specifying
doSegments=FALSE and doCalls=FALSE.
exportBins() also supports BED files.
exportBins() saves base pair positions in fixed notation instead
of scientific.madDiff()
from an S4 to an S3 method in version 0.9.4, released on
2014-05-23.getBinAnnotations() fixed after being broken by a change in
Bitbucket.Added exportBins() for exporting to a file.
Switch graphics in the vignette to PNG.
plot() honors user-specified values for xlab and xaxt.
plot() allows omission of labels for the standard deviation and
the number of data points.
Improve diagnostic messages.
smoothOutlierBins() correctly ignores bins filtered out.