CHANGES IN VERSION 1.19
-----------------------

SIGNIFICANT USER-VISIBLE CHANGES

    o qa(..., type="fastq") uses a sample of n=1000000 reads by
      default, rather than then entire file; use sample=FALSE to
      revert to previous behavior.

NEW FEATURES

    o encoding,FastqQuality and encoding,SFastqQuality provide a
      convenient map between letter encodings and their corresponding
      integer quality values.

    o filterFastq transforms one fastq file to another, removing reads
      or nucleotides via a user-specified
      function. trimEnds,character-method & friends use this for an
      easy way to remove low-quality base.

BUG FIXES

    o writeFastq successfully writes zero-length fastq files.

    o FastqStreamer / FastqSampler warn on incomplete (corrupt) files

CHANGES IN VERSION 1.17
-----------------------

SIGNIFICANT USER-VISIBLE CHANGES

  o FastqSampler can return records in the order encountered in the
    sampled file.

  o Increase to 10000 the number of reads examined for determining
    Fastq quality type

  o as(FastqQuality, "numeric") returns a vector of quality scores
    concatenated end to end (previously cycle to cycle), without
    padding to effective equal width

BUG FIXES

  o trimTails, successive=TRUE would return inconsistent results

  o FastqStreamer, FastqSampler parse fastq files created with '\r'

CHANGES IN VERSION 1.15
-----------------------

NEW FEATURES

    o FastqStreamer accepts IRanges for selecting input records

SIGNIFICANT USER-VISIBLE CHANGES

    o as(ShortReadQ, "matrix") now accepts ShortReadQ instances with
    heterogeneous widths, returning a matrix x[i, j] with NA values in
    when j > width()[i].

BUG FIXES

    o readAligned, type="BAM" correctly adds required 'what' elements

    o FastqSampler would only randomize first read; introduced 1.13.9
    2011-12-02, fixed 1.15.4 2012-04-25

    o report(qa, ...) no longer produces obviously confused base
    calls per cycle

    o FastqFileList would fail to initialize correctly from a
    character vector

CHANGES IN VERSION 1.13
-----------------------

SIGNIFICANT USER-VISIBLE CHANGES

    o FastqSampler is considerably faster

    o FastqSampler and FastqStreamer require explicit close() to avoid
    warnings about closing unused connections

BUG FIXES

    o qa reports on very large lanes would overflow alphabetFrequency

    o qa report scales adapaterContamination correctly

    o FastqSampler would rarely sample fewer than requested reads

    o FastqSampler supports outputs of >2^31 - 1 total nucleotides

    o readFastq parses records with 0 width

CHANGES IN VERSION 1.11
-----------------------

NEW FEATURES

    o trimTails to trim low quality trailing nucleotides

    o trimEnds to remove arbitrary (vectors of) letters from reads or
    qualities

    o FastqStreamer to iterate over a fastq file

    o FastqFile, FastqFileList to represent fastq files

SIGNIFICANT USER-VISIBLE CHANGES

    o writeFastq has argument full, default value FALSE, disabling
    printing of identifier a second time in '+' line

    o srapply requires that options(srapply_fapply="parallel") or
    options(srapply_fapply="Rmpi") to enable parallel processing via
    fapply

BUG FIXES

    o SolexaRealign, SolexaAlign, and SolexaResult transposed strand
    information

    o FastqSampler segfaulted on some files

    o writeFasta had a semi-documented argument mode; it is now
    documented and as a consequence dis-allows argument 'append' that
    would previously have been passed to underlying methods.

CHANGES IN VERSION 1.9
----------------------

NEW FEATURES

    o Support for HiSeq tile layout

    o Track reads passing filters, including across logical filter
    operations

CHANGES IN VERSION 1.7
----------------------

BUG FIXES

    o qa() represented the per-cycle quality scores incorrectly; this
    influenced qa[["perCycle"]][["quality"]][["Score"]], but not the
    qa report.

    o qa() for type="SolexaExport" transposed the 'aligned' and
    'filtered' labels on all elements of SolexaExportQA. Thanks
    Nicolas Delhomme for the report.

    o report() failed when each read was unique. Thanks Peng Yu for
    the report.

SIGNIFICANT USER-VISIBLE CHANGES

    o The perCycleQuality graph in the qa report now includes boxplots
    for all cycles instead of just the median value.

    o A depthOfCoverage graph has been added to the qa report for BAM,
    Bowtie, SolexaExport and SolexaRealign file types.

    o An adapterContamination measure has been added to the qa report for BAM,
    Bowtie, SolexaExport, SolexaRealign and Fastq file types.

    o srorder is now stable (the original order of identical is
    preservered).

NEW FEATURES

    o Add class BAMQA. qa() can now be called on BAM files.

    o The param argument in readAligned() and qa() for 
    type="BAM" can now be a single ScanBamParam object
    or a list of them.  

    o FastqSampler can be used to draw samples from a fastq file.

CHANGES IN VERSION 1.5
----------------------

SIGNIFICANT USER-VISIBLE CHANGES

    o levels(strand(aln)) is c("+", "-", "*") (was c("-", "+", "*"))

    o Add USE.NAMES argument to srapply, minimum length to (internal)
    function ..reduce.

NEW FEATURES

    o Optionally retrieve multiplex bar code, paired read number, and
    id from SolexaExport (contribution from Nicolas Delhomme)

    o renew() and renewable() provide an interface to updating
    ShortRead instances

    o srapply checks for and uses multicore

    o readIntensities supports Illumina RTA '.cif' / '.cnf' files

    o readAligned type="BAM" parses BAM files, extracting simple (no
    indel) cigars

BUG FIXES

    o readIntensities type="IparIntensity" correctly handles multiple
    tiles

CHANGES IN VERSION 1.3
----------------------

SIGNIFICANT USER-VISIBLE CHANGES

    o coverage,AlignedRead-method has a changed interface (shift/width
    rather than start/end) and default behavior (return value in
    genome coordinates, rather than minimal covered region).

    o readAligned,character-method, type="Bowtie" and readFastq return
    FastqQuality by default.

    o coverage,AlignedRead-method now returns an RleList

NEW FEATURES

    o qa reports from _realign.txt, MAQMap files

    o QualityScoreDNAStringSet coercion methods

    o qa type="character" now accepts a filter argument with value
    srFilter()

    o alphabetByCycle supports variable-width XStringSets

    o qa,ShortReadQ and qa,list methods for qa on existing objects

BUG FIXES

    o Parse .gz realign files

    o alphabetScore,FastqQuality-method shifted quality by +1


CHANGES IN VERSION 1.1
----------------------

SIGNIFICANT USER-VISIBLE CHANGES

    o 454 quality scores are returned as FastqQuality-encoded

    o For functions accepting dirPath, pattern to name files, allow
    dirPath to be a vector of file names when pattern is character().

    o width() on ShortRead and derived classes (including AlignedRead
    now returns a vector of widths, of length equal to the length of
    the object.

NEW FEATURES

    o Add Bowtie as a 'type' value for qa and report

    o Add dustyScore() and dustyFilter() to identify low-complexity
    regions

    o Parse _qseq files (to ShortReadQ or XDataFrame)

    o Parse IPAR image intensity files _int.txt.p, _nse.txt.p, 
    _pos.txt

    o Create HTML-based quality assessment reports

    o Add trimLRPatterns() for ShortRead and derived classes
    (ShortReadQ, AlignedRead).

    o Add narrow() for ShortRead, QualityScore, and derived classes.

    o Use append() to append two objects of the same ShortReadQ or
    QualityScore and derived classes together

    o writeFastq for classes derived from ShortReadQ

    o Input functions support .gz or text files.

    o readIntensity reads Solexa image intensity files into R,
    including information about lane, tile, x, and y coordinates of
    each read.

    o readPrb returns different types of objects, depending on the
    'as' argument of the readPrb,character-method.

    o readXStringSet gets arguments skip, nrows; argument order
    changed slightly

    o New built-in SRFilters positionFilter, uniqueFilter to select
    reads aligning to particular positions, or to select only unique
    instances of reads aligning to each position.
    
    o readAligned gains a Solexa _results parser (_results files are
    listed as 'intermediate' in the Solexa manual, and not a good
    end-point for analysis)

    o readAligned gains a Bowtie output parser

    o readAligned gains ability to parse MAQ 0.7 version binary files

BUG FIXES

    o readQual would fail to read 454 quality scores correctly when
    these spanned more than one line of input per read

    o coverage treated reads as 1 base longer than they were

    o FastqQuality got the quality encoding off by one in as(x,
    "matrix")

    o qa_solexa.Rnw incorrectly displayed read occurences when lanes
    were presented out-of-order (an unusual occurence)

    o readAligned SolexaAlign, etc., updated to parse 'chromsome' and
    'position', and 'strand' information correctly

    o readAligned MAQMapview failed for most chromosome labels

CHANGES IN VERSION 1.0
----------------------

SIGNIFICANT USER-VISIBLE CHANGES

    o SRFilter allows construction of filters that can be used to
    subset existing data objects, or filter incoming (readAligned, at
    the moment) objects.

    o readAligned for Solexa-based alignments return 'strand'
    information as factor with levels "-", "+", "*" (strand not
    relevant), NA (no strand information available).

    o srorder, srsort, srrank, and srduplicated for AligendRead class
    now sort based on chromosome, strand, position AND sread; previous
    behavior can be recovers by extracting the sequences
    srsort(sread(aln)), etc.

    o Functions using SolexaPath now search all relevant directories,
    e.g., in analysisPath, rather than the first

BUG FIXES

    o 'run' in eland_export files is correctly parsed as a factor

(start date: 29 September, 2008)