Changes in version 1.6.3
------------------------
- Fixed example broken in 1.6.2

Changes in version 1.6.2
------------------------
- Fixed logging issues in PureCN.R and NormalDB.R
- Bugfix in findBestNormal with pool and num.normals > 1 (voom was target length 
  normalized read counts provided, not raw counts)


Changes in version 1.6.1
------------------------
- Fixed crash when VCF contained variants with 0 coverage
- Fixed crash in calculatePowerDetectSomatic when fdr was 0
- Minor edits to "Quick Start" vignette


Changes in version 1.6.0
------------------------

- Lots of improvements to command line scripts
- Improved somatic vs. germline status calling
- Better mapping bias estimation and correction
- Better integration into existing copy number pipelines
- Support for cell lines
- New GC-normalization for smaller gene panels
- Added sub-clonal SNV state (SOMATIC.M0)
- Added beta-binomial model for noisy allelic fractions
- Polished plots, added new GC-normalization and volcano plots
- Better copy number normalization using multiple best normals
- Removed automatic curation, since the tuned likelihood model 
  of runAbsoluteCN was hard to beat
- More control over homozygous deletions (significant portion
  of wrong maximum likelihood solutions had many homozygous deletions)  
- Faster post.optimize=TRUE by not optimizing poor fits or unlikely solutions
- Automatic 50bp interval padding
- Tweaks to segmentationPSCBS
- seg.file can contain multiple samples
- Contamination rate estimation (experimental)
- Code cleanups (switch from inlinedocs to roxygen, from message/warn to
  futile.logger)


API CHANGES

- runAbsoluteCN output from PureCN 1.2 cannot be analyzed with PureCN 1.6 and 
  needs to be re-run. We hope to avoid this in the future.
- Renamed functions: readCoverageGatk to readCoverageFile since future versions
  will likely support additional third-party tools.
- Deprecated functions: createSNPBlacklist, getDiploid, autoCurateResults
- Defunct functions: createExonWeightFile
- Changed defaults: 
    - min.normals 4 (from 10) in setMappingBiasVcf
    - max.segments 300 (from 200) in runAbsoluteCN               
    - min.targeted.base 5 (from 4) in filterTargets
    - max.homozygous.loss now a double(2) vector, with first element
      specifying the maximum fraction of genome deleted (default 5%) and the 
      second value the maximum size of a homozygous loss (default 10mb).
    - prior somatic for variants in both dbSNP and COSMIC changed from 0.01
      and requiring 3 hits to 0.5 and requiring 4 hits.  
- Other minor changes:
    - Renamed some predictSomatic() output column names
    - Removed "beta.model" from "SNV.posterior" slot since model is now an
      option
    - Moved remove.off.target.snvs to filterVcfBasic
    - Moved normalDB from filterTargets to runAbsoluteCN, since it is now 
      used for more than target filtering
    - Dropped BED file support in calculateGCContentByInterval
      Instead provide support for GRanges
    - poolCoverage: w argument now used as provided, not 
        normalized so that w[1] is 1
    - Removed ... from runAbsoluteCN
    - min.coverage removed from segmentation function, since this is now
      done by filterTargets
    - Added centromeres to segmentation function
    - Replaced contamination.cutoff with contamination.range in filterVcfBasic 
    - Removed verbose from most functions, since messages are now controlled
      with futile.logger
    - Smoothing of log-ratios before segmentation now optionally done by 
      runAbsoluteCN, not segmentation function
    - setMappingBiasVcf now returns a list with elements bias (the old 
      return value) and pon.count, the number of hits in the PON
 
 
PLANNED FEATURES FOR 1.8

- Better sample summary statistics, like mutation burden, chromosomal 
  instability
- Better performance in low purity samples 
- Better performance in high purity samples with significant heterogeneity  
- LOH database
- Switch to S4 data structures (maybe)
- Whole dataset visualizations (maybe)
- Better support for known, small deletions and amplifications (e.g. EGFRvIII,
  MYC)
- Support for GATK4
- Better runtime performance by ignoring unlikely solutions early
