Changes in version 2.12.0 BACKWARDS-INCOMPATIBLE CHANGES - The range argument of the annotateCTSS function is renamed annot. - The removeSingletons option of clustering methods is removed and the default value of keepSingletonsAbove is set to 0, which keeps the standard behavior. - In cluster objects, the dominant CTSS score is now stored in the dominantCTSS object directly. - The clusterCTSS function is replaced by the new paraclu and distclu function. CTSS filtering is done beforehand with the new filterLowExpCTSS function. - Removed CTSSclusteringMethod() function and stop recording clustering method name. - Removed returnInterquantileWidth argument of functions and always return that width when quantile information is provided. Closes #114 - The CTSStagCountDA function is removed. - The dominant peak in TagClusters objects is now a GRanges object like in ConsensusClusters. - The custom method for tag clustering is removed. It was obsoleted by the newer CustomConsensusClusters function. - The exportToTrack function now exports scores of tag clusters and consensus clusters instead of setting them to zero. NEW FEATURES - Support the use of TxDB objects for annotating clusters. - The plotReverseCumulatives() function now uses ggplot2. Its main, legend, xlab, ylab, xlim and ylim arguments were removed as this can be controlled via ggplot2 functions. - New TSSlogo function wrapping the ggseqlogo package. - New distclu and paraclu functions that can run directly on CTSS objects. You can use them to test parameters before running the whole CAGEexp object through clusterCTSS. BUG FIXES - CTSS filtering now works correctly with threshold = 0, thresholdIsTpm = TRUE. - The importPublicData function was repaired for FANTOM samples. - Remove broken and obsolete customClusters method of clusterCTSS. Use CustomConsensusClusters() instead. Fixes #113. - Ensure consensus clusters can be exported as tracks with interquantile width information. Fixes #108, #70. OTHER CHANGES - Accelerated the computation of cumulative sums ~10×. - Singleton filtering is now done by the paraclu and distclu functions themeselves; .ctss_summary_for_clusters does not change the input clusters except for adding information. Changes in version 2.10.0 Changes in version 2.8.0 BUG FIXES - Correct quantile positions, which were shifted by one base. This bug may have been introduced in version 1.22 or later. - Ensure cluster objects are properly sorted. Fixes #79, introduced in version 2.6.0 and causing crashes or incorrect quantile calculations. - Apply fix for #77 (aggregateTagClusters losing TCs), which slipped out of 2.6.0 because of Git branch mixup. - Fix consensus cluster coordinates, where the maxDist padding was erroneously remaining in some parts of the computation. - Corrected on-the-fly cumulative sum computation for consensus clusters when sample = NULL. The bug was causing incorrectly short quantile ranges. - Force the cluster names to stay sorted, to avoid a bug desynchronising quantile information and genome coordinates. - Fix accidental deletion of the record of the tag clustering method (#96). NEW FEATURES - Allow URLs to files in getCTSS() (Fixes #50). - Accelerated the computation of cumulative sums and quantile position ~10×. - New resetCAGEexp() function. - New flagByUpstreamSequences() function. - The annotateCTSS and annotateConsensusClusters function gain a upstream and a downstream parameter to change the width of promoter regions. Changes in version 2.6.0 BUG FIXES - Re-enable the promoter shift functions (Fixes #4). - Compute dominant TSS information for consensus clusters (Fixes #1). NEW FEATURES - Inter-quantile width is also computed when no sample is selected. - Enhancer detection using a wrapper to the CAGEfighter package's function quickEnhancers(). Changes in version 2.4.0 Changes in version 2.2.0 BUG FIXES - Restore the correctSystematicG option in getCTSS(). See #61. - Restore object class consistency in if / else statement in private function bam2CTSS. Fixes #49. - Restore proper CTSS conversion from BAM files (bug introduced in v1.34.0) while fixing issue #36. - Ensure Tag Clusters have a Seqinfo. Fixes #63. Changes in version 2.0.0 BACKWARDS-INCOMPATIBLE CHANGES - The CAGEset class is removed. - Accessors using plain data.frame formats are removed. - Modifier functions return the object instead of silently modifying it in the global environment. Thus, you can use R pipes (|>). - Removed export function that unconditionally wrote files to the working directory, such as exportCTSStoBedGraph. As a replacement a new converter is provided, exportToBrowserTrack, that produces UCSCData objects that the user can then wite to files using the rtracklayer_ package. - Removed the extractExpressionClass function, as the information is easily accessible from within the CTSS or ConsensusClusters objects. NEW FEATURES - CTSSnormalizedTpmGR and CTSStagCountGR now accept all as a special sample name to return them all in a GRangesList object. - Plot interquantile width and expression clusters with ggplot2. BUG FIXES - Corrected the getExpressionProfiles function to accept CAGEexp objects. - Updated the exampleCAGEexp object to contain expression classes. - Restore paraclu support for CAGEexp objects. - Corrected a bug in aggregateTagClusters that was causing mislabelling of tag clusters (PR#42). - Prevent plotReverseCumulatives from crashing when values are not in range. (PR#43). Changes in version 1.34.0 BUG FIXES - Reform the CTSS class. New accessor: CTSS() (with no dot). - Correct a class error when loading BAM files. (Closes #36). - Use the BSgenome object from the main environment if available. Changes in version 1.32.0 BUG FIXES - Update end coordinates before start coordinates in the function .aggregateTagClustersGR(). This should stop triggering "'width(x)' cannot contain negative integers" errors. - Correct .make.consensus.clusters internal function. This should stop triggering "Consensus clusters must not overlap with each other" errors. - Allow empty CTSS.chr objects. - Correct plotInterquantileWidth() to really use consensus clusters when passed the argument clusters = "consensusClusters". - Fix failures on CAGEexp objects containing only one sample. Changes in version 1.30.0 BUG FIXES - Correct usage of && for vector comparison in pairs.DataFrame(). - Adjust to latest SummarizedExperiment MultiAssayExperiment versions. Changes in version 1.28.0 BACKWARDS-INCOMPATIBLE CHANGES - The CTSS objects are now based on the UnstitchedGPos class instead of the parent GPos class. Existing CAGEexp objects might give errors after package upgrade. NEW FEATURES - Depends on MultiAssayExperiment instead of previously importing it. No need to load it by hand (nor SummaryExperiment) anymore. BUG FIXES - Correct strand in .remove.added.G private function (PR#26). Changes in version 1.26.0 DOCUMENTATION - Split the "CAGE resources" documentation into a separate vignette. Changes in version 1.24.0 NEW FEATURES - plotAnnot(): the group argument now accepts formulas such as ~ a + b to indicate names of metadata column that will be pasted together to form a new group factor. BUG FIXES - Prevent mergeSamples() from producing colData that cause other functions to crash later when coercing to data.frame. - Repaired paraclu support for CAGEset objects. - normalizeTagCount() works again on CAGEset objects. - consensusClustersGR() reports expression score of the selected sample (instead of silently ignoring the sample argument and reporting expression sum on all the samples). Changes in version 1.22.0 BACKWARDS-INCOMPATIBLE CHANGES - The plotting functions send their output to the graphical device instead of writing it to a file. This makes their use more consistent with most plotting functions in R. NEW FEATURES - New CAGEexp class extending the MultiAssayExperiment class. It stores expression data more efficiently than CAGEset, and uses core Bioconductor types natively. For backwards compatibility it also support many of the original generic functions for CAGEset objects. Sample names in CAGEexp objects must by syntactically valid. Some methods are not yet available for CAGEexp object (see the vignette). Conversely some new functions will not be available for CAGEset objects. - New CTSS and TagClusters classes wrapping GPos and GRanges objects, for more type safety. - New functions for quality controls such as plotAnnot() or hanabiPlot(). See the CAGEexp vignette for details. - New removeStrandInvaders() function to count and remove strand-invasion artefacts (see Tang et al., 2013, doi:10.1093/nar/gks112). - Data export as DESeqDataSet object for DESeq2 with the new consensusClustersDESeq2() and GeneExpDESeq2() functions. - New bedctss format to load the FANTOM5 and FANTOM6 CAGE data. - Multicore parallelisation with BiocParallel instead of parallel. - New function sampleList() to help looping on samples with lapply(). - New plotCorrelation2() function, faster than plotCorrelation() because it is plain black and white. - Multicore loading of CTSS data in CAGEexp objects. OTHER CHANGES - Example data exampleCAGEexp, exampleCAGEexp and exampleZv9_annot are now is lazy-loaded. - NULL can be passed as genome name, to circumvent the requirement for a BSgenome object when actually not needing one. - In CAGEexp objects, expression quantile positions are given relative to the cluster start site. - For performance reasons, the positions of a quantile Q is now calculated as the position of the first base where cumulative expression is higher or equal to Q% of the total expression of a cluster. - plotReverseCumulatives(): - fitInRange accepts a value of NULL to turn off power law fitting. - New legend argument to remove legend when set to FALSE. - Axis range and labels can be modified with xlab/ylab and xlim/ylim. - Ladders steps start on the values instead of being centered on. - setColors(): allow lowercase in color names. BUG FIXES - Corrected a bug that was crashing CAGEset objects when loading more than one BAM file. - Load BAM as gapped alignment with readGAlignments() instead of scanBam(). Without this correction, TSS position on minus strand is incorrect in case of indels in the read. DOCUMENTATION UPDATES - Roxygen is used to generate the manual pages. - A new HTML vignette describes the CAGEexp class. Changes in version 1.18.1 - getCTSS() can now load data from large BAM files. Before this update, there were "negative length vectors" errors when filtering out low quality reads from datasets with multiple dozens of millions of sequences. Changes in version 1.18.0 - Added Charles Plessy as co-maintainer. - Remove warning by replacing deprecated ignoreSelf() with drop.self(). Changes in version 1.13.4 - 26/04/2016 updated the vignette. - 26/04/2016 added support for BED file CAGE tags input - included the new accepted value for inputFilesType="bed". - 24/04/2016 added support for paired-end BAM file CAGE data input (CAGEscan). Included the new accepted value for inputFilesType="bamPairedEnd". Changes in version 1.13.3 - 18/04/2016 updated the vignette. - 14/04/2016 modified the consensusClusters() function to allow extracting sample-specific information, such as CAGE signal, position of the dominant TSS and interquantile width in a specific sample (i.e. consensusClusters() now behaves analogously to tagClusters() function). - 10/03/2016 fixed the bandwidth of the smoothing kernel for plotting correlation density plots to always scale with the number of bins. - 10/03/2016 added new argument to plotCorrelation() function that allows controlling figure size in pixels. - 03/11/2015 fixed the format of exporting bed12 files for tag clusters, which contained overlapping blocks per record and caused problems when trying to convert bed to bigBed. - 03/11/2015 added an option of exporting CTSS bedGraph files to BigWig format - added a new argument to exportCTSStoBedGraph() function. Changes in version 1.13.0 Changes in version 1.11.0 Changes in version 1.10.2 - 11/10/2015 changed to selective import from data.table package to resolve the conflict regarding the shift function. Changes in version 1.10.1 - 15/05/2015 updated the vignette. - 15/05/2015 enabled changing xlim and ylim when plotting interquantile width distribution. - 13/05/2015 added mergeCAGEsets() function for merging two CAGEset objects. - 11/05/2015 defined as method for coercing a data.frame into a CAGEset object. Changes in version 1.9.3 - 06/02/2015 CAGEr package has now been published in NAR! If you find this package useful in your work, please cite: Haberle et al. (2015) CAGEr: precise TSS data retrieval and high-resolution promoterome mining for integrative analyses. Nucleic Acids Research 43(8):e51, doi:10.1093/nar/gkv054. Check citation("CAGEr") for more details. - 13/01/2015 fixed the error message in tagClusters() function for the case no qLow and qUp values are specified. Changes in version 1.9.2 - 03/12/2014 updated the vignette - fixed the references. Changes in version 1.9.1 - 01/12/2014 fixed the bug in importing data from ENCODEprojectCAGE data package using importPublicData() function when importing from multiple datasets that contain groups with the same name. Changes in version 1.9.0 Changes in version 1.7.2 - 01/06/2014 replaced the .RData files in the /data folder with the more compressed ones. Changes in version 1.7.1 - 30/05/2014 updated the vignette with the description of available public CAGE data resources and how to use them in CAGEr. - 25/05/2014 added a list of available human and mouse FANTOM5 samples as a dataset that can be loaded via call to .data(FANTOM5humanSamples) or data(FANTOM5mouseSamples). - 25/05/2014 modified the importPublicData() function so that it supports importing FANTOM5 data for human and mouse directly from FANTOM web resource, as well as data from several CAGE data packages. - 23/05/2014 optimized the creation of CTSS table across samples by performing merge on data.tables rather than data.frames - reading the data into CAGEset object from BAM or CTSS files is now ~10× faster. Changes in version 1.7.0 Changes in version 1.5.5 - 13/02/2014 fixed bug in function for plotting interquantile width (in cases of multiple pairs of low and high quantiles present it was sometime fetching multiple columns and stopping with an error). - 12/02/2014 fix typo in Details section of man page for exportToBed function. - 12/02/2014 fixed the bug in function for systematic correction of G addition bias (it was stopping with an error in a rare case when all reads beginning with a G and mapping to the genome are derived by trimming the G mismatch from upstream position). Changes in version 1.5.3 - 08/11/2013 fixed the bug in importPublicData function that caused problems when importing only one sample. - 08/11/2013 fixed error in BibTex file. Changes in version 1.5.2 - 06/11/2013 updated vignette and NEWS file. Changes in version 1.5.1 - 02/11/2013 moved vignette source to /vignettes folder. Changes in version 1.5.0 Changes in version 1.3.10 - 24/09/2013 optimized calculation of K-S p-values for shifting promoters using vectorized functions. - 23/09/2013 removed a .C call to the pKS2 function from stats package and replaced it by an R implementation of pKS2. Changes in version 1.2.9 - 23/08/2013 added consensusClustersTpm() function for retrieving matrix with TPM values for consensus clusters across all samples. Changes in version 1.2.8 - 22/08/2013 modified all functions to work on data from only one strand (some were previously giving error in such case). - 21/08/2013 fixed the bug in calling CTSSs from reads on minus strand when not correcting for G addition, i.e. when removeFirstG=FALSE. Changes in version 1.2.7 - 08/08/2013 added new argument to aggregateTagClusters() function (excludeSignalBelowThreshold) that controls which TCs will contribute to the total CAGE signal of the consensus cluster - only the ones above the threshold that are used initially to set the boundaries of the consensus cluster, or all TCs that overlap the resulting consensus clusters. - 07/08/2013 enabled retrieving interquantile width for tag clusters - added new returnInterquantileWidth argument to tagClusters() function. - 06/07/2013 enabled importing entire CTSS table with multiple columns of tag counts representing multiple samples/experiments - allowed setting imputFilesType to "CTSStable" when creating CAGEset object and extended getCTSS function accordingly to read the entire table at once and fill the object. Changes in version 1.2.6 - 22/07/2013 replaced multicore package dependency with parallel package (suggested by Bioconductor core team). - 21/07/2013 fixed the bug in using loaded genome when correcting for G nucleotide bias. - 20/07/2013 fixed plotReverseCumulatives to work with only one sample in the CAGEset. - 19/07/2013 changed naming of correlation plot files to distinguish CTSS from consensus clusters plot. Changes in version 1.2.5 - 28/06/2013 updated the vignette. Changes in version 1.2.4 - 16/06/2013 allowed plotting reverse cumulatives for normalized tag counts - added new argument to plotReverseCumulatives() function. - 15/06/2013 implemented calculating suggested referent power-law distribution and visualizing the distribution and its parameters on reverse cumulatives plot when calling plotReverseCumulatives() function. - 14/06/2013 allowed assigning user-specified colors to samples to be used in visualizations - added new setColors() function. - 13/06/2013 allowed plotting scatter plots and correlation for both individual TSSs (raw or normalized values) and consensus clusters - added new argument to plotCorrelation() function. Changes in version 1.2.3 - 12/06/2013 implemented plotting pairwise scatter plots of CAGE tag count and calculating correlation between samples - added new plotCorrelation() function. Changes in version 1.2.2 - 30/05/2013 implemented the algorithm for correcting 'G' nucleotide addition bias to CAGE tags described in Carninci et al., Nature Genetics 2006 - added a new option correctSystematicG to getCTSS function. Changes in version 1.2.1 - 28/05/2013 fixed the bug in reporting number of CTSSs in cluster when clustering CTSSs using custom clusters (for empty clusters it was returning that the number of CTSSs is 1, although the signal is 0) - 28/05/2013 fixed the bug in clustering CTSSs using custom clusters when run with multicore=FALSE (it was returning to many clusters - correct specified region but all of the regions on each chromosome, not only the specified one). - 22/05/2013 fixed the column classes in CTSS data.frame when reading the data from bam files (to match the classes when reading from ctss files). - 21/05/2013 updated the vignette with explanation on how to use custom build genomes. - 21/05/2013 allowed usage of custom build BSgenome packages by removing the check whether the specified genome is present in the available.genomes() from BSgenome package. Changes in version 1.2.0 - 02/05/2013 updated the vignette. - 02/05/2013 added a new feature in the plotReverseCumulatives function, so that the slope of the fitted power-law distribution in the user selected range of values is shown on the plot for each sample (helps to choose appropriate alpha parameter for normalization). - 02/05/2013 fixed a bug in selecting the range of tag count values for fitting power-law distribution in normalization. - 01/05/2013 added sample labels as names to library sizes vector. - 29/04/2013 implemented statistical testing (Kolmogorov-Smirnov test) for differential promoter usage based on cumulative sums of CAGE signal along the promoters (implemented within the scoreShift function). - 20/04/2013 added pass-through option none for method parameter in normalizeTagCount function to enable using raw tag counts in downstream steps (CTSS clustering, promoter width, etc.). - 19/04/2013 optimized scoreShift function to extract and process only the cumulative sums for samples being compared. - 19/04/2013 replaced lapply with a for loop in the scoreShift function to avoid invoking multicore within lapply. - 18/04/2013 fixed wrong error message in plotExpressionProfiles function that notifies about accepted values for what parameter. - 27/03/2013 added data.table to the list of dependencies and optimized various parts of the code to use data.table instead of data.frame.