Release 1.0.0

Major feature release with the following changes:

Major Changes

• Cross-building support moved from [2.11, 2.12] -> [2.12, 2.13]
• Support added for the high-performance Intel Inflator and Deflator for working with gzipped data
• Significant performance improvements to CallDuplexConsensusReads and the addition of multi-threaded calling
• A new 100% scala API for reading, writing and working with VCF files

Minor Changes

• Broken pipes while writing to stdout/stderr will print a concise error instead of a long stack trace
• Common option to fgbio.jar to set validation stringency when reading/writing SAM/BAM
• Minor fixes to HapCutToVcf
• UmiConsensusCaller and related tools now merge platform values in read groups case-insensitively

Release 0.8.1

Minor point release with a single new tool to sort FASTQ files by read name and number.

Release 0.8.0

Major release with the following changes:

• Major improvements to the pairwise Aligner class:
  • Significant performance improvements in the Aligner class for pairwise alignments
  • When aligning DNA sequences aligner will produce matches in CIGAR for matches between compatible IUPAC codes (e.g. R paired with A or G)
  • New method to produce all alignments above a score threshold from a pair of sequences
  • New interface to allow for custom gap scoring
• Added Sequences.revcomp() function that correctly reverse complements all IUPAC DNA/RNA codes
• Added method to Metric class to return an Iterator over a metrics file instead of reading the whole file into memory
• Io object now automatically supports bgzipped files with .bgz or .bgzip extensions
• Fixed bug in SamReader that would occasionally cause exceptions with overlapping query regions
• Updated to latest scala point version to create classes/JARs compatible with JDK 9 and 10 at runtime
• Added method to ExtractBasecallingParamsForPicard to enable easy access to unmatched BAM file path

Release 0.7.0

Release 0.7.0 introduces the following changes to existing tools:

• GroupReadsByUmi
  • check that the raw UMI tag is found foreach read (#406)
  • Fix log message in GroupReadsByUmi to be more accurate / less misleading (#436)
• DemuxFastqs: enable --quality-encoding to be used on the command line (#417)
• HapCutToVcf
  • fix ambiguous (IUPAC) reference bases on the fly #418)
  • add an option to skip indexing the output file (ex. when the input does not have CONTIG lines) #418)

In addition, the following new tools were added:

• FindSwitchbackReads: Tool to detect templates with strand-switch events in them (#438)

The following API changes were also introduced:

• FastqSource can handle read numbers > 2 (#408)
• Fixed writing and parsing of Double.Nan, Double.PositiveInfinity and Double.NegativeInfinity in Metric classes (#411)
• SamBuilder should accept missing bases and quals with a cigar (#424)
• Add message to require() call in Sample (#425)
• ReadStructure to allow and strip out whitespace within the read structure during parsing (#425)
• ProgressLogger.record should return if logging was triggered and a method to log the last record (#421)
• Bug fix: Metric.write was not closing its writer (#421)
• Adding a few useful methods to Sequences (#421)
• Metric now extends Commons Writer so we can use AsyncWriter on it (#437)
• Improve the error message when validating a sample shee. (#412)

Release 0.6.1

Bug fix release which resolves a problem introduced in a dependency that caused fgbio to be unable to read BAM files from stdin or named pipes. All users of 0.6.0 should upgrade to 0.6.1.

Release 0.6.0

Release 0.6.0 introduces the following changes to existing tools:

• ReviewConsensusVariants: output PASS when there are no filters on the variant; fix format of bases output
• MaskPrimers: improved usage documentation to make primer file format clearer

The following API changes were also introduced:

• Added constants to SamRecord for SAM/BAM related constant values
• NeedlemanWunchAligner renamed to Aligner (old name deprecated by still works)
  • Implemented Glocal (or semi-global) alignment mode
  • Impleemnted Local alignment mode
  • Fixed affine gap implementation
  • Fixed Alignment.subByQuery/subByTarget to correctly handle adjacent deletions
• In metrics files, ensure 0.0 always formats as 0 and not 0E0
• Updated how Rscript finds resources in the classpath to support local paths and absolute paths with and without leading slashes

Release 0.5.1

Release 0.5.1 is a minor bug-fix release and introduces the following changes:

• ExtractUmisFromBam
  • Improved error messaging
  • Fixed bug that prevented it from working when only one read per pair contained a UMI
• GroupReadsByUmi now adds the sub-sort SS tag to the header of BAMs produced
• CallMolecularConsensusReads and CallDuplexConensusReads attempt to detect the sort order of input data and will fail if the sort order is incompatible
• DemuxFastqs changed some output metrics from 32-bit Int to 64-bit Long to avoid overflows on NovaSeq data

Release 0.5.0

Release 0.5.0 introduces the following changes to existing tools:

• CallDuplexConsensusReads: Fixed a rare bug where the consensus base quality could be zero or one if the two strands' base qualities differ by two or less.
• FilterConsensusReads: Fix for bug where duplex reads formed from raw reads from a single strand only could be incorrectly filtered.
• CorrectUmis: Now stores the original UMI sequences in the OX tag upon correction.
• DemuxFastqs: Bug fix to correct quality scores in output BAM files
• ClipOverlappingReads: Removed previously deprecated tool. Use ClipBam instead.
• ClipBam:
  • Now optionally outputs metrics about clipping present in reads before and after execution.
  • New option to "upgrade" clipping, e.g. replace existing soft-clipping with hard-clipping

Changes to APIs were as follows:

• Various deprecated methods were removed this release.
• Metric formatting now prints smaller Doubles in scientific notation, and the formatting is generally more efficient.
• NeedlemanWunchAligner gained a Glocal alignment mode for aligning all of a query sequence to a sub-region of a target sequence

Release 0.4.0

Release 0.4.0 introduces the following changes to existing tools:

• CallDuplexConsensusReads
  • The single strand consensus bases and quals for each duplex consensus read are output into tags on the duplex consensus read
  • Added option to output consensus reads that are formed from only a single strand
• FilterConsensusReads
  • New option to filter out reads with low mean base quality
  • New option to filter out reads whose minimum depth is too low
  • New option to filter duplex consensus reads where the single strand consensuses disagree
  • New optional tags will store the the single-strand consensus bases and qualities for duplex consensus reads.
• DemuxFastqs
  • will no longer output /1 and /2 on read names when running in Illumina standards mode
  • fixed a bug causing an exception when the sample barcode is found in multiple reads (ex. i5 and i7)
• ErrorRateByReadPosition - fixed bug that resulted in C>G errors being counted as A>G errors
• GroupReadsByUmi
  • Reads with UMIs with Ns in them are now rejected
  • Log messages added with counts of reads filtered out by reason
  • Memory usage improvements when grouping reads at very, very high depth.
  • Supports enforcing a minimum UMI length and partial UMIs except for the paired strategy (duplex sequencing).

Finally, changes to various APIs were as follows:

• Method in Bams to sort records by tag, or by a function applied to a tag
• Improve speed of Metric.read for loading large numbers of rows from metrics files
• Changed SamSource to extend IterableView instead of Iterable so that map(), filter(), etc. return lazy views
• Fixed a bug where the specified temporary directory was not being used for sorting.
• Added a BinomialDistribution class implemented using unlimited precision decimal math which is slower, but allows computation of cumulative probabilities where other implementations overflow or underflow

Release 0.3.0

Release 0.3.0 introduces the following changes to existing tools:

• ClipBam - The --overlapping-reads option was not being used internally and is deprecated in favor of --clip-overlapping-reads. This caused overlapping reads to always be clipped.
• CollectDuplexSeqMetrics - Added the optional output of duplex-umi frequencies with DuplexUmiMetrics.
• DemuxFastqs - The default output sort order is changed from Unsorted to Queryname. Add an option --illumina-standards to output file names using Illumina naming conventions. Tuned the amount of memory used, especially for a large # of samples (>96).
• CallDuplexConsensusReads - Do not except when we find potential collisions in duplex molecules, instead, do not generate a consensus read.
• FilterBam - adding a few more filters.
• Added a global parameter for log-level.

In addition, the following new tools were added:

• CollectErccMetrics - This will collect metrics for analyzing ERCC spike-ins in
  RNA-Seq experiments for dose response but not fold-change
  response.

Finally, changes to various APIs were as follows:

• ReferenceSetBuilder - Moved to the testing packages for use in projects that extend fgbio.
• Alignment - Added subByQuery() and subByTarget() methods to Alignment.