3-26-2013:  1.4
        * added gBGC functions phastBias, bgc.nucleotide.tests, and
	  classify.muts.bgc
        * changed all "alt.mod" to "ls.mod" (lineage-specific)
	* added nuc.text (and options nuc.text.pos, nuc.text.col) options to
	  plot.msa and as.track.msa; this is for plotting additional text
	  along the top or bottom of an alignment.
        * fixed bug in alt.mod.tm (was ignoriong rate.matrix argument, which
	  has been removed)
        * modified overlap.feat to return list of two gffs when
	  get.fragments==TRUE (one containing fragments and another containing
	  features that selected the fragments)
	* modified sort.feat to sort first by seqname, then by position
	* added flatten.feat which merges adjacent features of the same type
	* fixed bug in as.track.msa which (refseq arg was ignored)
        * changed definition of selection in lineage-specific models to be
	          the sum of mod$selection + mod$altmod$selection
	* fixed bug causing segfault in concat.msa
	* added max.EM.its argument to phyloFit
	* added optim.rphast function
	* updates to feat object; now it belongs to classes "feat" and
	  "data.frame"
	* update to guess.format.msa; now it opens up files and guesses format
	  based on contents by default.
	* added freq3x4.msa to calculate codon frequencies of an alignment using
	  3x4 model
	* added site.model flag to tree model object which indicates whether
	  a tree model represents a Nielsen-Yang site model
	* added function setup.branch.site.tm which takes a tree model and
	  returns a tree model representing a branch-site model with a
	  specified foreground branch.  The site-model can optionally
	  include extra site categories with bgc on the foreground branch.
	* changed tagval and tagval.feat to work with either GFF 2 or 3 
	  specification
12-22-2010 : 1.3
	* improvements to memory handler, which previously slowed down rphast
	  by 33%.  Now there is no significant increase in run-time, at least
	  for the code in rphast/tests.  It also no longer causes occasional
	  segfaults when code is interrupted, since protection now occurs
	  during the on.exit function instead of only when the function
	  finishes cleanly.  We also protect all external pointer objects,
	  rather than only the ones the user might see, which also cleans up
	  potential segfaults and makes writing new RPHAST functions more 
	  foolproof.
	* fixed bug in phyloFit when optimizing codon models with the
	  selection parameter; now rate matrix is scaled before applying
	  selection parameter.
	* fixed bug in plot.msa where alignment plot was placed wrongly when
	  ploted with several tracks.
	* changed phyloP to always return 1-based coordinates
	* added function mod.backgd.tm which modifies background frequency of
	  tree model while maintaining reversibility
	* added function translate.msa
	* added function col.expected.subs.msa
	* use %g notation (instead of %f) to print branch lengths of trees;
	  this avoids some loss of precision when passing trees to/from R/C.
11-10-2010: 1.2 released 
	* added memory handler so that memory allocated in C is no longer
	  lost due to function interrupt (or failure to explicitly free it)
	* added support for alternate lineage-specific models within
	  the tree model object
	* added support for codon models (HKY_CODON, REV_CODON, SSREV_CODON),
	  also codon.clean.msa function
	* to support lineage-specific models, added support for labels in
	  phylogenetic trees which are denoted by the pound sign ie,
	  "((human #primate, chimp #primate)#primate, mouse)"
	* many changes to phyloFit; allow lineage-specific models, arbitrary
	  parameter boundaries, hold specified boundaries constant.  Some
	  of the default options are now different from command-line phyloFit.
	* added plot.msa and as.track.msa functions which plot alignment; only
	  currently works if alignment fits in the plotting window.  Also
	  added coord.range.msa function to support this.
	* added plot.rate.matrix (and plot.tm and plot.altmodel.tm)
	* added set.rate.matrix.tm and get.rate.matrix.params.tm which set 
	  a rate matrix based on substitution model parameters, or get the 
	  substitution model parameters describing a rate matrix, respectively.
	* added summary.tree function which returns a data frame with various
	  information about each branch
	* fixed bug in split.by.feature.msa in which offset of resulting msas
	  was one higher than it should have been
	* changed several function names to increase consistency in naming 
	  conventions:
	    addUTRs.feat -> add.UTRs.feat
	    addIntrons.feat -> add.introns.feat
	    addSignals.feat -> add.signals.feat
	    fixStartStop.feat -> fix.start.stop.feat
	    validFormatStr.msa -> is.format.msa
	    isSubstMod.tm -> is.subst.mod.tm
	    branchlen.tree -> branchlength.tree
	    wig.track -> as.track.wig
	    feat.track -> as.track.feat
	    gene.track -> as.track.feat(is.gene=TRUE)
	* added pairwise.diff.msa function which returns pairwise differences
	  per site between pairs of sequences in an alignment
	* fixed some minor bugs in documentation (multiple notes in functions,
	  help("[.msa") not working previously)
	* Switched order of arguments in write.feat to (feature, file) for 
	  consistency with all other write functions.
	* added functions postprob.msa, expected.subs.msa, and 
	  total.expected.subs.msa
	* allow ref.idx argument in phyloP to refer to features object
	* removed region.bounds argument from coverage.feat; now requires at 
	  least one features object have not=FALSE

2010-09-17: 1.1 released
	* updated DESCRIPTION
	* updates to Makevars and configure.ac to resolve error compiling 
          on Solaris
	* fixed bug causing segmentation fault in gff.c

2010-09-15: 1.0 released