Changes in version 1.2.3
------------------------
- Critical fix for bug in setMappingBiasVcf. Somatic mutations were not
  excluded.  This bug results in a too large mapping bias estimate and very
  wrong calls in hyper-mutated samples. Only affects matched tumor/normal
  pairs.
- Un-deprecated segmentationPSCBS for panels with copy number tiling probes.


Changes in version 1.2.2
------------------------
- Fix for rare crash in very heterogeneous samples 
- Clarified interval padding in the vignette.
- Un-deprecated segmentationPSCBS. 


Changes in version 1.2.1
------------------------

- Fix for floating point overflow when iterations was set to a high value
  (https://support.bioconductor.org/p/88053/)
- Fix for bug in callAlterationsFromSegmentation (resulting in wrong
  gene.mean, gene.max, gene.min values)
- Check for overlapping intervals in readCoverageGatk  

  
Changes in version 1.2.0
------------------------

Focus on PureCN 1.2 was to dramatically improve user-friendliness:

- GATK requirement was dropped: PureCN now comes with functionality to
  generate coverage data when GATK is not available.
- Experimental support for VCFs generated by other callers than MuTect 1.1.7
- Experimental automatic curation of results.
- Output plots were polished.
- Both numerical (NCBI-style) and non-numerical (UCSC-style) chromosome names 
  (1 vs. chr1) supported.
- Better integration into existing copy number pipelines.
- Automatic COSMIC annotation.
- Thorough checks of input data, resulting in fewer crashes with unhelpful 
  error messages.
- Documentation improvements.


OTHER MAJOR ENHANCEMENTS

- More thorough initial grid search should minimize cases where purity/ploidy
  is wrong because the solution was not considered.
- More robust and accurate likelihood model that provides robust estimates of 
  minor segment copy numbers and more accurate LOH.  
- Support for SNVs outside the target interval file
  (remove.off.target.snvs=FALSE).   
- Improved segmentation and log-ratio normalization.  
- Support for non-human samples.
- Support for 100% pure samples (matched normal mode only).
- Faster SNV fitting.
- Experimental support for correcting non-reference mapping bias.


API CHANGES

- New functions: autoCurateResults, bootstrapResults, 
    calculateBamCoverageByInterval, calculateGCContentByInterval,
    calculateLogRatio, calculatePowerDetectSomatic, callAlterations, 
    callAlterationsFromSegmentation, callLOH, filterTargets, getDiploid, 
    getSexFromVcf, setMappingBiasVcf
- Deprecated functions: segmentationPSCBS
- Renamed functions: createExonWeightFile to createTargetWeights
- Renamed function arguments: 
    exon.weight.file to target.weight.file
    gatk.normal.file to normal.coverage.file
    gatk.tumor.file to tumor.coverage.file
    coverage.cutoff to min.coverage
- Changed defaults:
    findFocal: made defaults less stringent, now 3Mb (instead of 2Mb) and
        minimum copy number of 5 (callAlterations still uses 6 as default).
    runAbsoluteCN(genome): no default anymore (instead of 'hg19')    


OTHER NEW FEATURES

- Functions to calculate power to detect mono-clonal and sub-clonal somatic
  mutations (Carter et al., Nature Biotech, 2012) were added.


PLANNED FEATURES FOR PURECN 1.4 (BIOCONDUCTOR 3.5)

- Support for indels.
- Better runtime performance by ignoring unlikely solutions early.  
- Better support for known, small deletions and amplifications (e.g. EGFRvIII,
  MYC)
- Better support for pool of normals.
- Switch to S4 data structures (maybe).
- Whole dataset visualizations (maybe).
