NEWS

methylKit 1.3.7
--------------
IMPROVEMENTS AND BUG FIXES
* if methylDiff object contains NA in a qvalue/meth.diff column 
  the getMethylDiff function returns an error. This is now fixed:
  https://github.com/al2na/methylKit/issues/83
* Changed how the methylBase version of regionCounts handle missing values.
  Before, a missing value made the whole region/tile missing. Now, a missing
  value is treated as a site with zero coverage. Contributed by Karl Nordström:
  https://github.com/al2na/methylKit/pull/77

methylKit 1.3.6
--------------
* changes to methSeg() function: update function description 
  by beeing more explicit about the sorting, 
  sort(x,ignore.strand=TRUE) is now called once if needed, 
  update tests 

methylKit 1.3.5
--------------
IMPROVEMENTS AND BUG FIXES
* fix problem with methSeg() error occurring when GRanges is not sorted by position

methylKit 1.3.4
--------------
IMPROVEMENTS AND BUG FIXES
* fix problem with assocComp() on methylBaseDB 

methylKit 1.3.3
--------------
IMPROVEMENTS AND BUG FIXES
* changes to calculateDiffMeth() function: add a message explaining 
  the calculation procedure based on either two or more treatment groups, 
  fixed bugs that prohibited the use of treatment groups other than 0/1 
  in two group case, allow for more than two groups, update tests
* changes in methSeg() function: update function documentation to 
  inform that provided Granges has to be sorted, 
  add check if granges is sorted and contains at least one meta.col

methylKit 1.3.2
--------------
IMPROVEMENTS AND BUG FIXES
* quick fix for long filename issue: in unite() and calculateMethylDiff() 
  instead of concatenated sample_ids, filename will be "methylBase" 
  concatenated with either 13-char random string or provided suffix
* updated vignette: added two short FAQs

methylKit 1.3.1
--------------
IMPROVEMENTS AND BUG FIXES
* fix conversion rate error in methCall() 

methylKit 1.1.8
--------------
IMPROVEMENTS AND BUG FIXES
* fix methSeg error when only one segment is returned from fastseg,
  add case handling for methSeg2bed
* check for user interruption in methCall to enable stop in execution
* changes to select() function: check for out-of-bound indices to 
  prevent downstream errors
  
methylKit 1.1.7
--------------
IMPROVEMENTS AND BUG FIXES
* fix methCall segementation fault, added tests and test files from bismark
* fixed missing p.value at coercion of methylBase to GRanges
* changes to the pool() function: save.db=TRUE for methylBaseDB by default, 
  differing lengths of given sample.ids and unique treatment lead to error, 
  added tests
* fix osx related error when reading gzipped files with methRead
* change deprecated function names in test files
* fixed bug with dataSim() function updated the manual

methylKit 1.1.6
--------------
IMPROVEMENTS AND BUG FIXES
* fixed a bug where tileMethylCounts() function did not work with
  small chromosomes/scaffolds with few bases covered.

methylKit 1.1.5
--------------
IMPROVEMENTS AND BUG FIXES
* fixes missing error messages during methRead : https://github.com/al2na/methylKit/pull/57

methylKit 1.1.4
--------------
IMPROVEMENTS AND BUG FIXES
* merging tabix files is fixed: https://github.com/al2na/methylKit/pull/56

methylKit 1.1.3
--------------
IMPROVEMENTS AND BUG FIXES
* Typos in the vignette are fixed, thanks to Marcin Kosinski 
  
methylKit 1.1.2
--------------
IMPROVEMENTS AND BUG FIXES
* Bug fixes in SAM file reading process. If there were more than one
  header line there were problems in reading. Now this is fixed.
  https://github.com/al2na/methylKit/pull/51
  
methylKit 1.1.1
--------------
IMPROVEMENTS AND BUG FIXES
* Fisher's exact test now works as described in the manual. It is automatically
  applied in calculateDiffMeth() when there are only two groups with one 
  replicate each.
* During logistic regression modeling, the samples without counts are removed 
  from the model but the same filtering is not applied for covariates 
  data.frame, which can cause errors if min.per.group argument is used. 
  Now this is fixed: https://github.com/al2na/methylKit/issues/50
  
  
methylKit 0.99.4
--------------
IMPROVEMENTS AND BUG FIXES
* all annotation functions are now removed and are available    
  through genomation package. See the vignette for examples. Most
  functions have similar names and functionality in genomation. 
  The users need to convert methylKit objects to GRanges using 
  as(methylKit.obj,"GRanges") before they can use genomation 
  functions.


methylKit 0.99.3
--------------
IMPROVEMENTS AND BUG FIXES
* Fixed a bug in processBismark() C++ function where BAM files are should be treated as 0-based.
  Now  this is fixed.
* Bug in calculateDiffMethDSS is fixed (https://github.com/al2na/methylKit/issues/49)


methylKit 0.99.2
--------------
IMPROVEMENTS AND BUG FIXES

* Fixed a bug in methRead() introduced after the addition of mincov argument.
  the bug occurred only reading files legacy text files that have CpGs
  with coverage below 10
* Changes to vignette for better description of the tests.
* Compiler error that occurs in older compilers are fixed via this PR 
  https://github.com/al2na/methylKit/pull/43


methylKit 0.99.1
--------------
IMPROVEMENTS AND BUG FIXES

* mostly changes to meet BioCcheck()  requirements and reccomendations
* C++ code compiles on windows now, regex requirement is no longer there.
 

methylKit 0.9.6
--------------
IMPROVEMENTS AND BUG FIXES

* changes to following function names:
  read() to methRead()  
  read.bismark to processBismarkAln()
  adjust.methylC() to adjustMethylC() 
  get.methylDiff() to getMethylDiff()
  annotate.WithFeature() to annotateWithFeature()
  annotate.WithFeature.Flank() to annotateWithFeatureFlank()
  annotate.WithGenicParts() to annotateWithGenicParts()
  read.bed() to readBed()
  read.feature.flank() to readFeatureFlank()
  read.transcript.features() to readTranscriptFeatures()
* Improved documentation for methRead() (old read())
* Now, bismark cytosine report and coverage files can be read using methRead()
  pipeline argument. see ?methRead
* Ported the Perl script for methylation base calling to C/C++ via Rcpp. 
  Contributed by Alexander Gosdschan.
* methRead() uses data.table::fread() to read files faster.
* methRead() has a new argument mincov, which sets the minimum number of
  reads that needs to cover a base. Positions with coverage below this 
  number are discarded.

NEW FUNCTIONS AND FEATURES

* new function methSeg() can segment methylation (methylRaw objects) and
  differential methylation (methylDiff objects) profiles to segments.
  Associated function methSeg2bed() creates BED files from segments.
  see ?methSeg. A test is added to check this in R CMD check.
  Contributed by Arsene Wabo and Alexander Gosdschan.

* new tabix based classes methylRawDB, methylRawListDB, methylBaseDB, 
  methylDiffDB and respective methods implemented. Tests are updated to 
  check proper function in R CMD check. Contributed by Alexander Gosdschan.

* calculateDiffMeth() now supports basic overdispersion correction and multiple 
  methods for pvalue correction. The function also now handles covariates 
  such as age,sex etc. A test is added to check this in R CMD check. 
  Contributed by Adrian Bierling.

* New function calculateDiffMethDSS() is using beta-binomial model from DSS 
  package to calculate differential methylation. Contributed by 
  Dhruva Chandramohan with modifications from Altuna Akalin. This is a modified
  version of the function from DSS package so that it can work with methylKit
  objects.

* dataSim creates a methylBase object with simulated methylation data. 
  A test is added to check this in R CMD check. Contributed by Adrian Bierling.

methylKit 0.9.5
--------------
IMPROVEMENTS AND BUG FIXES

* travis CI build shield added


methylKit 0.9.4
--------------
IMPROVEMENTS AND BUG FIXES

* tileMethylCounts now works on methylBase objects, affected by BioC 3.0 upgrade.
  A test is added to check this in R CMD check.


methylKit 0.9.3
--------------
IMPROVEMENTS AND BUG FIXES

* compatibility with BioC 3.0. Multiple BioC functions moved to other packages
  within BioC,which broke some code and caused installation issues. 
  Now this is fixed.
* data.table::merge is now stable when all=TRUE. Removed my(altuna) version of data.table::merge
  the code base.


methylKit 0.9.2.5
--------------
IMPROVEMENTS AND BUG FIXES

* calculateDiffMeth slim=FALSE argument works correctly when there are multiple
  samples per group.


methylKit 0.9.2.4
--------------
IMPROVEMENTS AND BUG FIXES

* install_github() now works correctly. Removed blank lines at the end of
  DESCRIPTION file.

* calculateDiffMeth and tileMethylCounts now works correctly,
 typos were introduced in code with 0.9.2.2. Now these are fixed


methylKit 0.9.2.3
--------------
IMPROVEMENTS AND BUG FIXES

* regionCounts() bug occurring when the first argument is a class of methylBase is fixed.
  The bug introduced an additional column to the resulting methylBase object.
  https://groups.google.com/forum/#!topic/methylkit_discussion/p19K-pgavAI


methylKit 0.9.2.2
--------------
IMPROVEMENTS AND BUG FIXES

* unite() issues when destrand=FALSE is resolved. The issue appeared due to numeric vs.
  integer conflict when merging data sets using chr,start and end locations.
* typo in calculateDiffMeth() argument "weighted.mean" is fixed.
* unite() is faster when destrand=TRUE, due to improvements in internal function
  .CpG.dinuc.unify()
* regionCounts() strand.aware argument now works correctly. This bug had no effect
  on tileMethylCounts. With the default arguments, strand aware setting was not
  in effect at all, so every region was treated as strandless.
* Conversion to GRanges objects now removes seqlevels(chromosome names)
  that are not used. Not removing these could cause problems in regionCounts()
  function and functions depending on regionCounts().
* read() function in some cases was treating strand columns of the flat CpG
  files as logical if strand had F or R letters and most of the first CpGs were
  having F strand (Forward strand). Now this is fixed.

methylKit 0.9.2.1
--------------
IMPROVEMENTS AND BUG FIXES

* Fixes a bug in getMethylationStats() function that generates incorrect label numbers when the user overrides the default number
  of breaks for the histogram. Contributed by Bonnie Barrilleaux.

methylKit 0.9.2
--------------
IMPROVEMENTS AND BUG FIXES

* A bug introduced with 0.9.1 is fixed. The bug occured when unite() function
is used with destrand=TRUE argument. It returned less number of CpGs then it was
supposed to, although returned CpGs had correct methylation and coverage values.


methylKit 0.9.1
--------------

NEW FUNCTIONS AND FEATURES

* objects takes less memory and they are reorganized. updateMethObject() updates
  the objects from previous versions the latest version.

* New batch effect control functions are implemented. You can control if certain principal
  components are associated with batch effects and remove those components from your
  data. see ?assocComp amd ?removeComp. In addition, if you have corrected for the
  batch effects via other methods, you can reconstruct a corrected methylBase object (
  see ?reconstruct). Check the "batch effects" section in the vignette.

IMPROVEMENTS AND BUG FIXES

* unite() function takes less time due to use of data.table::merge

* fixes a bug appeared with R 3.02, where getting exon and intron coordinates
  from BED12 files produced an error.

* tested with R 3.2 and matching bioconductor packages

* data(methylKit) has the new objects



methylKit 0.5.7
--------------


IMPROVEMENTS AND BUG FIXES

* tested with R 3.0 and matching bioconductor packages

* deprecated matchMatrix import from IRanges was causing a problem with package installation. v0.5.7 fixes that problem.

* Now there is no need for "chr" string in BED files when reading them as annotation files. Some assemblies do not have the "chr"" string in their chromosome names.


methylKit 0.5.6
--------------

NEW FUNCTIONS AND FEATURES

* new arguments for clusterSamples() and PCASamples() functions are added. With
  the new options "sd.filter","sd.threshold","filterByQuantile" are added. These
  options help finetune how low variation bases/regions are discarded prior to
  clustering or PCA. See ?PCASamples and ?clusterSamples() for details on the
  new options

* FAQ section added to the vignette

* show methods added for each class. Now, typing the variable name containing the
  object will display concise information about the contents of the object.

* Subsetting objects via "[" notation is now enabled. You can subset rows of the
  objects and it will return a new object rather than just a data frame.


IMPROVEMENTS AND BUG FIXES

* tileMethylCounts() error  is fixed. Error occurred when tilling sparsely
  covered small chromosomes like chrM in human RRBS data.

* read.bismark() can deal with Bismark output with Bowtie2. Bowtie2 can put gaps
  in the alignment, now read.bismark() can deal with those gaps when parsing the
  SAM file.

* Coverage columns are coerced to integer when reading generic methylation per
  base files. BSMAP scripts can produce a methylation ratio file where coverages
  (or "effective CT counts" as they are called)  are not always integers, which
  causes a problem in the downstream analysis. Now, these non-integer columns
  are rounded to nearest integer while reading.
  See http://zvfak.blogspot.com/2012/10/how-to-read-bsmap-methylation-ratio.html
  for example usage of this functionality.



methylKit 0.5.5
--------------

IMPROVEMENTS AND BUG FIXES

* Differential methylation percentage calculation bug fixed. The bug occurred when "min.per.group" argument used in unite() function and when "weighted.mean=TRUE" in calculateDiffMeth() function.

* plotTargetAnnotation bug is fixed. Bug occured when "precedence" argument set to FALSE.


methylKit 0.5.4
--------------

IMPROVEMENTS AND BUG FIXES

* Examples added to help pages

* Changes to DESCRIPTION for complying with the bioconductor guidelines

* unused "cor" option removed from PCASamples() function

* some irrelevant functions are not exported (i.e they are not public) anymore.

* getContext() looks for the correct slot name now


methylKit 0.5.3
--------------

NEW FUNCTIONS AND FEATURES

* new function adjust.methylC() can be used to adjust measured 5mC levels by measured 5hmC levels


IMPROVEMENTS AND BUG FIXES

* pool() function bug fixed where one of the groups have one sample the pool function was not returning correct values

* calculateMethDiff() function option "SLIM" is now working. If set to TRUE SLIM method for q-value calculation will be used. If FALSE, p.adjust with method="BH" option will be used for P-value correction.

* read.bismark() function now works correctly under Windows.

* read.bismark() bug occurring when the reads are paired and overlapping is fixed now.

* read.bismark() bug occurring when the alignment is done in non-directional manner is fixed now. However, illumina sequencing protocols are directional and you are unlikely to have encountered this error if you aligned your sequences in a directional manner.

* getCorrelation() function has a new option called "method" can take the value of "spearman","person" or "kendall"

methylKit 0.5.2
--------------

NEW FUNCTIONS AND FEATURES

* new function pool() sums up coverage, numCs and numTs values within each group so one representative sample for each group will be created in a new methylBase object.

* new function normalizeCoverage()  normalizes coverage and associated number of Cs and number of Ts values between samples using a scaling factor derived from the ratio between mean or median of coverage distributions of samples.


methylKit 0.5.1
--------------

IMPROVEMENTS AND BUG FIXES

* calculateDiffMeth() can now deal with differential methylation calculations where one group has multiple samples but the other one has only one.


methylKit 0.5
--------------

NEW FUNCTIONS AND FEATURES

* new function reorganize() can be used to reorganize methylRawList and methylBase objects to create new objects from subset of samples or can be used to change the order of samples and treatment vector.

* new function bedgraph() can output UCSC bedgraph files for methylRaw and methylDiff objects

* new function percMethylation() extracts percent methylation values from a methylBase object and returns a matrix

* unite() now can merge bases/regions that are not covered by all samples by setting "min.per.group" option

* PCASamples() has new options "transpose" and "sd.threshold". Now, one can do PCA analysis on the transposed % methylation values. "sd.threshold" is for removing rows with small variation prior to PCA analysis  .see ?PCASamples for details

IMPROVEMENTS AND BUG FIXES

* regionCounts() bug that appeared with data.table 1.8.0 is fixed

* unite() bug that appeared with R 2.15 is fixed

* calculateDiffMeth() can deal with NA values introduced by setting "min.per.replicate" option of unite() function

* PCASamples() now uses "prcomp" function to do the PCA analysis

* external annotation data (cpgi.hg18.bed.txt and refseq.hg18.bed.txt) that comes with the package is now only a subset of full datasets. Do not use them for your own analysis, download the complete annotation data from UCSC or other sources in BED format.

methylKit 0.4.1
--------------

NEW FUNCTIONS AND FEATURES

* new function select() can be used to subset methylRaw, methylBase and methylDiff objects to create new objects with a subset of methylation data useful if you want to use only a particular portion of methylation events on the genome.

IMPROVEMENTS AND BUG FIXES

* read.bismark bug fixed, now it should be able to save methylation call files with no problem

methylKit 0.4
--------------

NEW FUNCTIONS AND FEATURES

* New read.bismark() function can read directly from a sorted SAM file output by Bismark aligner, the function can save methylation calls as text files or read them as methylRaw object to be used in analysis

* calculateDiffMeth() can do all differential methylation calculations using multiple-cores if 'num.cores' is set to an integer denoting how many cores should be used.

* getCoverageStats() & getMethylationStats() have a new option called 'labels', if set to FALSE, no labels will be drawn on top of the histogram bars.

IMPROVEMENTS AND BUG FIXES

* cov.bases option in tileMethylCounts() now works

* methylRaw,methylBase and methylDiff objects have a new slot 'resolution', which designates
  whether methylation information is base-pair resolution or regional resolution. allowed values 'base' or 'region'

* getCoverageStats() & getMethylationStats() now print the sample ids and methylation context automatically in the
  title when plotting

* getCoverageStats() & getMethylationStats() takes extra options to be passed to hist() function

* destrand option of unite() function will be over-ridden when methylRawList to be united contains regions rather than bases

* unite() function checks if supplied elements of methylRawList object have the same context,assembly and resolution before
  uniting data sets.

* validity function checking the format of the data for a methylRaw object is implemented

* clusterSamples() and PCASamples() take methylation context information automatically and use it in plot titles

methylKit 0.3.1
--------------

IMPROVEMENTS AND BUG FIXES

* syntax error when fisher.test applied at calculateMethDiff() is fixed.