NEWS

methylKit 1.27.2
--------------
IMPROVEMENTS AND BUG FIXES
* fix R-Devel build error:
  - issues was caused by S4Vectors::isEmpty
  - fixed by replacing S4Vectors::isEmpty with is.null


methylKit 1.27.1
--------------
IMPROVEMENTS AND BUG FIXES
* fix R cmd check warnings:
  - fix "Undocumented arguments in documentation object" by 
  documenting ‘...’, ‘treatment’ arguments for methylRawList and 
  methylRawListDB constructors
  - methCall: replace variable length char with vector of chars to fix the 
  warning: "variable length arrays are a C99 feature [-Wvla-extension]"
  - methCall: simplify the parsing of cigar string and replace 
  calls to deprecated std::sprintf with std::snprintf
* add test for processBismarkAln to check that reading sam and bam 
returns the same object for the same content


methylKit 1.17.5
--------------
IMPROVEMENTS AND BUG FIXES
* fread.gzipped: skip header rows in tabix file to fix https://github.com/al2na/methylKit/issues/226



methylKit 1.17.4
--------------
IMPROVEMENTS AND BUG FIXES
* .setMethylDBNames: correct possible methylDBclass from "methylDB" 
  to "methylRawDB"
* fread.gzipped: disable skipping of decompression in fread.gzipped
  - can cause serious issues as investigated in 
  https://groups.google.com/g/methylkit_discussion/c/UruFjvX89B4/m/vV2Qnd8NEAAJ 
  and explained in https://github.com/al2na/methylKit/issues/222


methylKit 1.17.3
--------------
IMPROVEMENTS AND BUG FIXES
* export methylRawListDB and methylRawList constructors

methylKit 1.17.2
--------------
IMPROVEMENTS AND BUG FIXES
* update vignette:
  - add note about use of system.file function
  - reduce number of rows, when showing row ordering
  - add faq section about merging methylRaw into methylRawList update vignette

methylKit 1.17.1
--------------
IMPROVEMENTS AND BUG FIXES
* readmethylDB: check if file exists before trying to read
* loading tabix files: 
  - improve reading tabix files with header
  - update tests for loading tabix files

methylKit 1.15.3
--------------
NEW FUNCTIONS AND FEATURES
* add verbosity argument to methCall to allow for 
  complete silencing of output, this will be set by 
  processBismarkAln which gained 'verbose' argument

IMPROVEMENTS AND BUG FIXES
* update R requirement to (>= 3.5.0)
* processBismarkAln:
  - add verbose argument
  - update tests
  - update docs
* joinsegmentneighbours: 
  - Catch case when only one groups is there
* regionCounts:
  - update docs: mention that the output will be sorted by default
  - Fixes #195
* vignette: 
  - add faq paragraph about regionCount to the vignette explaining 
  why we order input region
  - update paragraph about methyldackel when to use 
  it instead of processBismarkAln 
  - Fixes #162
* add extensive test for bedgraph export
* update and export tests for getMethylationStats
* add extensive tests for getCoverageStats
* add extensive tests for normalizCoverage
* update checks for dataSim:
  - if sample ids given check if length(sample.ids) == # replicates 
* update tests for getMethylDiff
* unite destranding: 
  - make sure high numbers are not used in scientific notation
  - fixes #129
  - switch to data.table based destranding as it is faster and 
  more memory efficient for larger number of regions
* methseg: 
  - catch error when joined segments produce only single range
  - Fixes #195
* calculateDiffmeth: 
  - check if output is equal for memory and tabix file
  - Fixes #189

methylKit 1.15.2
--------------
IMPROVEMENTS AND BUG FIXES
* check.dbdir: catch if path equals root ("/") dir
* methRead: 
  - add missing context argument for call 
  to .procBismarkCytosineReport
  - fixes https://github.com/al2na/methylKit/issues/198

methylKit 1.13.4
--------------
IMPROVEMENTS AND BUG FIXES
* set stringAsFactors to FALSE 
  - changed when reading data and tabix files
  - this solves class missmatches downstream
  - update methylKit data with stringsAsFactors off
* methRead:
  - filter for given context when pipeline is cytosinereport
  - order data when processing with methread, this ensures that 
    methylRaw and methylRawDB contain exact same data
  - add additional test for methread, check whether methread 
    with or without tabix produces same data when processing 
    bismark coverage and cytosine report
* unite:
  - make sure .CpG.dinuc.unifyOld equals .CpG.dinuc.unify
  - add test to check that destranding result is same 
    for methylBase and methylBaseDB

methylKit 1.13.3
--------------
IMPROVEMENTS AND BUG FIXES
* getMethylDiff: 
  - fixes bugs in method for methylDiffDB
  - suffix will be actual type now, was "type" before
  - removed whitespaces in filename,
  fixes https://github.com/al2na/methylKit/issues/171
  - add extensive test for getmethylDiff function
* applyTbxByChunk:
  - Allow optional writing of colnames and rownames if 
  returntype is text. Find details in 
  https://github.com/al2na/methylKit/issues/169
* percMethylation:
  - return rownames for methylDB when rowids is TRUE,
  fixes https://github.com/al2na/methylKit/issues/170
  - add extensive tests for percMethylation
* reorganize:
  - fix bug in method for methylBaseDB introduced with new tabix header
  - add extensive testing for reorganize
* assocComp:
  - add further checks for proper annotation
* removeComp:
  - set default comp to NULL, to trigger earlier fail
  - fix bug method for methylBaseDB introduced with new tabix header
* reconstruct:
  - update checks for percent methylation matrix
  - fix bug with fread 
* add extensive tests for batch control functions
* diffMethPerChr: 
  - prevent from plotting when no chrom left
  - add extensive tests for diffMethPerChr
  
  

methylKit 1.13.2
--------------
IMPROVEMENTS AND BUG FIXES
* fread.gzipped: 
  - add pre-checks wheter file exists and is not empty
  - switch to R.utils::gunzip instead of building shell   
    command manually 
* tileMethylCounts: remove orphan warning triggered by   
  incomplete temporary methylRaw object, 
  fixes https://github.com/al2na/methylKit/issues/175
* update vignette: 
  - add mincov argument to methread
  - add paragraph to tiling section about reducing 
  coverage threshold + update example code
* update docs:
  - rebuild docs with more recent version of roxygen
  - exclude examples from internal function
*  increase version requirement for tabix files with header
  
  
    

methylKit 1.13.1
--------------
IMPROVEMENTS AND BUG FIXES
* fix issues related to build checking: 
  - logical vectors of size longer than one
  - docs for methseg contain line longer than 100 chars
* fix methSeg2bed error when seg.mean=0 and scores=NA by 
  replacing score with zero if average seg.mean is zero
  https://github.com/al2na/methylKit/issues/123
* update vignette:
  - explain how to get wig/bigwig file from methylKit
    https://github.com/al2na/methylKit/issues/157
  - explain how to deal with soft clipping in processBismarkAln
    https://github.com/al2na/methylKit/issues/159
  - add section about exporting and loading tabix objects 
    starting from version 1.13.1

methylKit 1.11.1
--------------
NEW FUNCTIONS AND FEATURES
* add functions to read and write a header to tabix files: 
  this allows for easier sharing of methylDB objects, 
  as the metadata is included in the tabix file
* new function readMethylDB: function reads the header from a given tabix file 
  and loads it into corresponding methylDB object


methylKit 1.9.4
--------------
NEW FUNCTIONS AND FEATURES
* calculateDiffMeth: introduce method "midPval" for fast testing of 
  different methylation if there is one sample per group, 
  contributed by @zellerivo 

IMPROVEMENTS AND BUG FIXES
* calculateDiffMeth: 
  - fix rare case issue with invalid p-values on calculateDiffMeth call, 
    that caused "qvalue" p-value adjustment to fail 
    https://github.com/al2na/methylKit/issues/90
  - let user choose between new "midPval" or older "fast.fisher" implemtation 
    for statistical test used to determine the methylation differences
    https://github.com/al2na/methylKit/issues/96
  - introduce p-value lookup table to reduce multiple testing of 
    identical configurations 
    https://github.com/al2na/methylKit/issues/96
  - reduce code duplication and overhead by introducing S3 functions 
    to be called by S4 methods
* update vignette with section about usage of methylKit for analysis of 
  affinity- and array-based methylation assays
* methSeg: if object with multiple chromosomes is used, remove chroms with 
  less than two ranges 
* unite: fix issue where digits were written in scientific notation 
  https://github.com/al2na/methylKit/issues/130
* getCorrelation: fix issue where getCorrelation produces matrix with NA
  when using "min.per.group" https://github.com/al2na/methylKit/issues/137



methylKit 1.9.3
--------------
IMPROVEMENTS AND BUG FIXES
* getCorrelation: reduce code duplication and overhead by 
  introducing S3 functions to be called by S4 methods
* fread.gzipped: rewrite building of shell command with 
  system2 to mimic R.Utils::gunzip
* improve validity checks for methylDB objects
* pool: check length of treatment and number of sample is same
* headTabix: use fread.gzipped only when user wants to get more than 1e6 rows
* test-1-read: check if methylRawDB is same when directory is given or not,
   i.e. test wether directory affects object
* methylRaw: check for content in tabix file, break if not
* calculateDiffMeth: dealing with case of full model not being different
  from reduced model; print warning and set NaN to 1
* pool: check length of treatment and number of sample is same
* improve validity checks for methylDB objects

methylKit 1.9.2
--------------
IMPROVEMENTS AND BUG FIXES
* fread.gzipped: update function to check compression based on file content 
  and allow to decompress .bgz files
* headTabix: use fread.gzipped directly on the tabix file
  fixes issue https://github.com/al2na/methylKit/issues/141

methylKit 1.9.1
--------------
IMPROVEMENTS AND BUG FIXES
* add methylRawListDB constructor
  fixes https://github.com/al2na/methylKit/issues/134
* update validation functions for methylRaw/methylRawDB: check for single sample id
* unite: write S3 functions for class methylRawList and methylRawListDB
  Fixes https://github.com/al2na/methylKit/issues/132


methylKit 1.7.10
--------------
IMPROVEMENTS AND BUG FIXES
* update man: update object and function descriptions to resolve 
  the 'file link treated as topic' warning under windows 

methylKit 1.7.9
--------------
IMPROVEMENTS AND BUG FIXES
* bug fix: make internal function .checkdbdir more error proof

methylKit 1.7.8
--------------
IMPROVEMENTS AND BUG FIXES
* internal applyTbxByOverlap(): did not work correctly for return.type set to
  "data.frame" or"data.table", due to missing argument.
  https://github.com/al2na/methylKit/issues/131

methylKit 1.7.7
--------------
IMPROVEMENTS AND BUG FIXES
* unite(): skip checking if destranded tabix file exist 

methylKit 1.7.6
--------------
NEW FUNCTIONS AND FEATURES
* methSeg(): introduce parameter `initialize.on.subset` to subset data
  for initialization of mixture modeling; update description; add tests

methylKit 1.7.4
--------------
IMPROVEMENTS AND BUG FIXES
* update link to test-file for methSeg() function


methylKit 1.7.3
--------------
IMPROVEMENTS AND BUG FIXES
* selectByOverlap(): update description; refine method signatures to 
  only support GRanges as range argument; update NAMESPACE to import 
  subjectHits() method from S4Vectors 


methylKit 1.7.2
--------------
NEW FUNCTIONS AND FEATURES

* New constructor method methylRawList() can be used to combine 
  list of methylRaw objects into a methylRawList

IMPROVEMENTS AND BUG FIXES
* fix bug in methSeg: when joinSegments was activated, 
  diagnostic plot would always be plotted
* fixes in selectByOverlap() function: update description, to show that any 
  methylKit object (tabix or not) can be used with it ;
  fixed broken method after @subjectHits was not available anymore 
* fix error in .checkTabixFileExists function, that lead to overwriting of files
  
  

methylKit 1.5.3
--------------
IMPROVEMENTS AND BUG FIXES
* resolve absolute paths for dbdir argument
* initial check if output tabix file already exists and if yes rename output file

methylKit 1.5.2
--------------
IMPROVEMENTS AND BUG FIXES
* updated links to vignette and presentations in README 
* fixed missspelling in show method of methylRawList object
* fixed getSampleID method for methylDiff object
* correct a not working code snippet in the vignette

methylKit 1.3.8
--------------
IMPROVEMENTS AND BUG FIXES
* changes to methCall(): more verbose output, reduce -Wunsigned warnings,
  check for fragmented chromosome order, write conversion stats to extra file
* add Tests to check for chromosome disorder 


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.