PWMEnrich package

Version 4.5.1:
    o Convert log(P-values) back to P-values for human using a chi-sq distribution

Version 4:
    o New algorithm for human backgrounds
    o New function: toPWM() that takes both PFMs and PPMs

Version 3.5:
    o After further testing revert back to PWMEnrich 2.x group P-value algorithm
    o Introduced group sorting by top motifs

Version 3.1.4 (devel branch only):
    o New way of estimating P-value for groups of sequences. Note this will produce different P-values for groups of sequences than PWMEnrich 2.x !

Version 2.4.4:
	o Vignette update and fix naming of columns in the motif enrichment report

Version 2.4.2:
    o Improve promoter selection for human and mouse genomes (duplicates are now disregarded)    

Version 2.4.0:
    o Major update with more functions and small bugfixes
    o Added sequenceReport() and groupReport() for easier report generation
    o Visualise motif scores along a sequence with plotMotifScores()
    o Creation of empirical CDFs for motif scores
    o Almost complete rewrite of the vignette to emphasize the main use cases
    o Converted documenation to roxygen2

Version 2.3.2:
    o Subsetting functions for backgrounds from Diego Diez

Version 2.3.1:
    o Fix a bug with plotTopMotifsSequence() with calling an unknown function 
    o Implement group.only for all background, not only pval in motifEnrichment()
    o New default to plotMultipleMotifs() so the margins are better
    
Version 2.2:
    o Bioconductor 2.12 release version (same as 2.0.0)    

Version 2.0.0:
    o General cleanup of the code with various small optimisations
    o A FASTA file name is now also taken as input to motifEnrichment()
    o The output of motifEnrichment() is now wrapped into a class MotifEnrichmentResults
      that provides a number of convenience methods for common tasks like ranking and
      plotting motifs
    o Functions makeBackground() and getBackgroundFrequencies() can now take BSgenome 
      objects as input. Thanks to Diego Diez for suggesting this and providing the code. 
    o Another version of motifScores() has been implemented that requires large amounts
      of memory, but is at least 2 times faster than the old motifScores() implementation.
      Use a new option useBigMemoryPWMEnrich() to switch to this implementation.   
    o PFMtoPWM now accepts a new parameter seq.count so that MotifDb motifs that are 
      expressed as probabilities instead of frequencies can be easily used.  
    
Version 1.3.0:
    o Bioconductor 2.11 release version    

Version 1.0.2:
    o Use core package parallel instead of doMC

Version 1.0.1:
    o Fix a typo in test cases and remove doMC as build dependency

Version 1.0.0:
    o Initial release of the package