BFAST -- History of visible changes.
See LICENSE for details
See COPYING for copying conditions.
Please send bfast bug reports to bfast-help@lists.sourceforge.net
Version 0.6.4d
* adding '-z' option in 'bfast postprocess' to choose a random best scoring
alignment when used with '-z 3'.
* MD tag in the SAM output modified to agree with 'samtools calmd'.
Version 0.6.4c
* bug fixed encountering '.' base characters. Added new tests in 'make check'.
Version 0.6.4b
* bug fixed in postprocess for color space where an the color space error
string was malformed, hitting an assertion.
Version 0.6.4a
* 'postprocess' now supports using pairing information to better choose
alignments.
* multi-threading re-implemented in 'match', 'localalign', and 'postprocess'.
* output format of BAF files changed to reduce file size, thus speeding up
'localalign'
* bfast.submit.pl now can clean up files no longer need to make the final BAM
file.
* some small boundary conditions fixed in 'localalign' that would cause a
crash.
* command line parsing now standardized with GNU getopt.
* mapping quality was incorrectly calculated in some boundary cases,
defaulting to the 255 value.
* MAF and GFF output removed.
* pairing/mirroring options in 'localalign' removed.
* ill2fastq.pl now outputs the qualities from sequence.txt files assuming
PHRED+64 input.
Version 0.6.3c
* NM tag now fixed to include indels as edits.
* MD tag was not considering leading matches when the read included
an indel.
Version 0.6.3b
* Added the 'bfast easyalign' command that will run bfast with the
default parameters ('bfast match', 'bfast localalign', and
'bfast postprocess').
* Additional mapping quality bug found when only two alignments are
present.
Version 0.6.3a
* The mate's reference name is now set to "=" when it matches
the current reads reference name when outputting to the SAM format.
* All characters are ignored after the leading whitespace when reading
in the names of the contigs in the reference FASTA file. Previously,
whitespaces where converted to '_'.
* Adding for reading .gz and .bz2 files in solid2fastq (C-version).
Also, the ability to write to .gz and .bz2 files is supported.
* Adding in the ability to read .gz and .bz2 files in 'bfast match'.
* Fixed bug related to calculating the mapping quality. This may
explain the high occurrence of 255 mapping qualities in the SAM output.
* Fixed a bug in the calculating the NH tag when using -a 3 in
'bfast postprocess'.
Version 0.6.2a
* Adding more optional tag fields to SAM output, most notably the
MD field. The supported fields are detailed in the bfast book.
* Adding multi-threading to "bfast postprocess"
* Changed qseq2fastq.pl to ill2fastq.pl to support Illumina both
QSEQ and SEQUENCE files.
* Optimized the binary search within the index search; should
be faster.
* Adding in an option to "bfast match" that will load all main
or secondary indexes into memory to avoid storing temporary files.
This is particularly useful on high-memory machines.
* Fixing a bug that did not properly save all constraints. This
reduced performance previously.
* Fixed behaviour differences between when the index(es) are
split and when they are one piece.
* solid2fastq (C-version) now can read in bzip2 compressed
files using the "-j" option. This is experimental.
Version 0.6.1
* fixing bugs related to the reorganization of the code.
Version 0.6.0
* complete re-organization of the bfast code and commands
* supports automatic splitting of the indexes to allow efficient
low-memory execution (>4GB for the human genome).
* bfast reference and index file names are now based off of the
input reference FASTA file.
* the match, localalign, and postprocess command support the
pipe-and-filter model (stdin and stdout).
* local alignment now supports a seed-constrained version, which
uses information from the lookups to constrain the local
alignment. This adds a speed improvement as well as accuracy
improvement.
Version 0.5.7
* bfast.submit.pl supports the -s and -e options in BFAST.
Version 0.5.6
* balign now is able to only align a specific subset of reads from the
matches file. This is similar to bmatches and should be especially
useful if the index search (bmatches) is faster than the local
alignment (balign), which is usually the case for ABI SOLiD data.
* Both the secondary indexes list and offsets files are now optional in
bmatches.
* Added bfast.submit.pl PERL script for scheduling alignment jobs on SGE
and PBS clusters.
Version 0.5.5
* Adding support for input queueing, so that a given number of reads are
loaded into memory then analyzed instead of one-by-one.
Version 0.5.4
* Bug in the SAM header and setting the unpaired reads incorrectly.
* SAM output will now skip triple/quad/... end reads instead of quitting.
* Adding support for all alignments to be carried forward. This is so
if a SAM file is generated, all alignments, even those that are unmapped,
are present in the SAM file. This affects bmatches, balign, and bpostprocess.
* qseq2fastq.pl now works with Illumina pipeline 1.4
Version 0.5.3
* Bug converting SAM format fixed where unmapped ends of a paired end
read where not being annotated correctly (if the end was the first or
second).
* Added C implementation of solid2fastq.pl for faster fastq generation.
* Fixing bug in qseq2fastq.pl that did not complement but only reversed
the seconde end of a paired end read.
Version 0.5.2
* Fixing bug when searching secondary indexes (introduced in 0.5.0
when zlib use started).
Version 0.5.1
* Fixed a bug when trying to read in secondary indexes in bmatches.
* Updated make check for creating the distribution.
* Truncating the keysize now works in bmatches.
* Updated the default scoring matrix so that it does not need to be
specified.
* Updating gzreads and gzwrites for custom 64-bit support.
* Updating the maximum number of matches behaviour in
* Adding in some new helper functions to calculate the power or sensitivity
of the alignments.
Version 0.5.0
* All output files now use the zlib for compression. This includes brg,
bif, bmf, and baf files.
* Removed brepair and breport as they are now obsolete.
Version 0.4.7
* Updated output file naming in bmfconvert.
* Updated bug in balign regarding retrieving the substitutions scores.
Version 0.4.6
* Simplified the scoring matrix. Now only match/mismatch is required for
input.
* Updated bug in color space local aligner if the color match score was
not zero. Also enforcing all color match and color mismatch scores
respectively to be the same to allow for an optimization in the algorithm.
Version 0.4.5
* Moved mapping quality calculation to balign. BAF files have an extra
field specifying the mapping quality. For 0.4.1-0.4.4 BAF versions, use
fixbaf to convert them to 0.4.5 versions.
* Removed bfix - obsolete.
Version 0.4.4
* Fixed bugs when outputting to SAM format.
* Added bmf conversion to fastq.
* Fixed bug in RGRanges, where pallindromic sequences were being removed
from one strand.
* Added new utility brg2fasta.
* Added ability in bmatches to use masks that are prefixes of the mask (-k).
* Now can output a very rudimentary mapping quality in bpostprocess.
* Added a scripts folder to convert raw base/color calls to BFAST fastq
format.
Version 0.4.3
* Added an option to bpostprocess where all "best scoring" alignments
can be returned.
* Fixed bug where aligning perfectly did not retain repeat masker
information in the reference.
Version 0.4.2
* Added support for SAM 0.1.2 output.
* Added new utility brepair, which tries to align unpaired reads. I
would not recommend using it since it does not in my analysis help
too much in practice and is inefficient.
Version 0.4.1
* Adding support for quality scores. The input reads now must be in
fastq or multi-end fastq format.
* Adding support for multi end reads.
* The names of Output files of bmatches, balign, and subsequent
programs are changed.
Version 0.3.4
* Updated balign, bmfconvert, bafconvert, and bsort output file names.
* Added gff output in bpostprocess and bafconvert.
* Fixed bug in bpostprocess when specifying the maximum number of
nt errors.
* bsort now performs a threaded sort if the maximum number of entries
is large.
* removed some options to do with choosing paired end reads in
bpostprocess, since they are easily misinterpreted.
* added a uniqueness score option in bpostprocess, which will replace
the alignment score with a ratio of the chosen alignment score versus
all other alignments' scores.
Version 0.3.3
* A quick fix to remove a memory leak that can causes crashes
when aligning nt space data.
Version 0.3.2
* Updating how we report color errors in .baf files.
* Added a new utility: bmergesorted that merges two files that
have been sorted by bsort.
* Updated bsort to print more information while sorting. Also,
updated bsort to sort bpostprocess unpaired baf output.
* Update berrordistribution and bpairedenddist to accept any
number of input files.
* Changed bgeterrordistribution to berrordistribution.
* Updated bpostprocess to output unpaired reads when one end
has ambiguity.
* Updated local alignment optimization in balign (-b).
* Updated bsort to take in the maximum number of entries for
memory management.
* Fixed bug in balign where alignment could not handle switching
directly from an insertion to a deletion (or vice versa).
Version 0.3.1
* Found bug in read simulator.
* Fixing more bugs to do with indels.
Version 0.2.6
* Fixed a bug in breport that caused memory leaks.
* Added support to output unpaired reads in bpostprocess.
* Fixed a bug in balign where the position of indels were
being reported incorrectly for NT space. This was as a result
of incorporating color space code incorrectly.
Version 0.2.5
* bpairedenddist now outputs to a file instead of stdout.
* Added a new utility balignsim which tests the accuracy of
various local alignment algorithms implemented by balign.
* Fixed bug in balign when considering indels for color space.
Version 0.2.4
* Added a .sh script to update the tests since the reference
genome files and index files have their version numbers embedded.
* Updated color space tests; there was a bug in reporting the
reverse strand.
* Updated balign "Validatin outputID" message.
* Moved from using mktemp to mkstemp, which means no more mktemp
warnings.
* Added a new utility, bsort, which sorts a bfast reported file.
* Fixed off by two error for reverse strand in color space for
bmatches.
* Fixed bug with paired end bmatches file as input to bevalsim.
* Fixed bug with calling insertions on the reverse strand. The
offset was off by the two times the length of the insertion.
Version 0.2.3
* Updated bfix to include packed.
* Updated bug when reading a bfast reference genome file.
* Added "make check" functionality for regression testing.
* Fixed warning in Mac OS X complaining of unused argp structures.
* Fixed bug in Mac OS X where -r argument in bpostprocess caused
an error.
* Updated names of files generated by bmatches and balign.
Version 0.2.2
* Fixed major bug that was introduced in 0.1.10 where matches to
the reverse strand were being erroneously included.
* Added check when making the indexes that the hash width is less
than or equal to the key-size.
Version 0.2.1
* Major version change.
* Added option in balign to allow local alignment to only consider
mismatches and color errors.
* Removed memory leak in balign.
* begenerate reads now outputes a read number in the read name.
* Halved the required storage size of the hash in the index.
* Added butil utility bfix which will convert 0.2.1 < reference
genome and index files to the current version.
Version 0.1.13
* Updated bug in RGIndex where reverse read could have n's.
Version 0.1.12
* bmatches: now option -w can specify which strand to match.
* bevalsim now takes in any bmf or baf file.
* Optimized part of bmatches procedure. Both forward and reverse
matches wil count towards maxKeyMatches (-K).
* Updated bpreprocess help message.
* Updated bindexhist and bindexdist for color space.
* Fixed bug in bgeneratereads for color space.
Version 0.1.11
* bpostprocess now prints out MAF that complies with UCSC standards.
Version 0.1.10
* bmatches supports only matching to the forward strand. This
facilitates aligning methylated sequences.
* bpreprocess bug fixed when using an exons file.
* bevalsim now requires knowing paired end info and the original
reads file generated by bgeneratereads.
* fixed bug in bgeneratereads and bevalsim for paired end data.
* Fixed bugs in AlignEntryRead where the contig name was not being
read properly.
Version 0.1.9
* Updated bmf output to properly display when the maximum number
of matches has been reached.
* bmfconvert does not require paired end argument.
* Now supports trimming whitespace including carriage return on the
ends of the reads, mainly to support reads made on Windows machines.
* Updated butil binaries.
Version 0.1.8
* Updated to support Ns in the reads; will still ignore "."s.
* Adjusting bounds bug with -e option fixed. Would not default to
the correct end position.
* Updated command line parsing to avoid memory leaks.
Version 0.1.7
* Update bpostprocess for paired end support: now tested.
* Added support for filtering each end by score for paired end reads.
* Removed extra "." in inversions file (bpostprocess).
* Updated btestindexes against small bugs and added optimizations.
* Added bgeneratedistribution to calculate error distributions.
Version 0.1.6
* bgeneratereads was not producing SNPs and errors properly.
* bmatches now supresses timing output when -t is not specified.
* bpreprocess memory leaks fixed.
Version 0.1.5
* Fixed bug in bpreprocess where the program crashed right before terminating.
Version 0.1.4
* Updated bmatches output file name for the filtered reads.
* Updated bpostprocess output file names.
* Recoded btestindexes to support finding indexes for color space.
Version 0.1.3
* Improved speed for index creation while specifying an exon-like structure.
Version 0.1.2
* Added support for index creation while specifying an exon-like structure.
Version 0.1.1
* First alpha public release.
-------------------------------------------------------
See LICENSE for details
