samtools/bcftools 1.4

on GitHub

Release 1.4 (13 March 2017)

Two new commands - mpileup and csq:

• The mpileup command has been imported from samtools to bcftools. The
  reasoning behind this is that bcftools calling is intimately tied to mpileup
  and any changes to one, often requires changes to the other. Only the
  genotype likelihood (BCF output) part of mpileup has moved to bcftools,
  while the textual pileup output remains in samtools. The BCF output option
  in samtools mpileup will likely be removed in a release or two or when
  changes to bcftools call are incompatible with the old mpileup output.

  The basic mpileup functionality remains unchanged as do most of the command
  line options, but there are some differences and new features that one
  should be aware of:

  • The option samtools mpileup -t, --output-tags changed to bcftools mpileup -a, --annotate to avoid conflict with the -t, --targets
    option common across other bcftools commands.

  • -O, --output-BP and -s, --output-MQ are no longer used as they are
    only for textual pipelup output, which is not included in bcftools mpileup. -O short option reassigned to --output-type and -s
    reassigned to --samples for consistency with other bcftools commands.

  • -g, --BCF, -v, --VCF, and -u, --uncompressed options from
    samtools mpileup are no longer used, being replaced by the
    -O, --output-type option common to other bcftools commands.

  • The -f, --fasta-ref option is now required by default to help avoid user
    errors. Can be diabled using --no-reference.

  • The option -d, --depth .. max per-file depth now behaves as expected
    and according to the documentation, and prints a meaningful diagnostics.

  • The -S, --samples-file can be used to rename samples on the fly. See man
    page for details.

  • The -G, --read-groups functionality has been extended to allow
    reassignment, grouping and exclusion of readgroups. See man page for
    details.

  • The -l, --positions replaced by the -t, --targets and
    -T, --targets-file options to be consistent with other bcftools
    commands.

  • gVCF output is supported. Per-sample gVCFs created by mpileup can be
    merged using bcftools merge --gvcf.

  • Can generate mpileup output on multiple (indexed) regions using the
    -r, --regions and -R, --regions-file options. In samtools, one
    was restricted to a single region with the -r, --region option.

  • Several speedups thanks to @jkbonfield (cf3a55a).

• csq: New command for haplotype-aware variant consequence calling.
  See man page and paper.

Updates, improvements and bugfixes for many other commands:

• annotate: --collapse option added. --mark-sites now works with
  VCF files rather than just tab-delimited files. Now possible to annotate
  a subset of samples from tab file, not just VCF file (#469). Bugfixes (#428).

• call: New option -F, --prior-freqs to take advantage of prior knowledge
  of population allele frequencies. Improved calculation of the QUAL score
  particularly for REF sites (#449, 7c56870). PLs>=256 allowed in
  call -m. Bugfixes (#436).

• concat --naive now works with vcf.gz in addition to bcf files.

• consensus: handle variants overlapping region boundaries (#400).

• convert: gvcf2vcf support for mpileup and GATK. new --sex option to
  assign sex to be used in certain output types (#500). Large speedup of
  --hapsample and --haplegendsample (e8e369b) especially with --threads
  option enabled. Bugfixes (#460).

• cnv: improvements to output (be8b378).

• filter: bugfixes (#406).

• gtcheck: improved cross-check mode (#441).

• index can now specify the path to the output index file. Also, gains the
  --threads option.

• merge: Large overhaul of merge command including support for merging
  gVCF files created by bcftools mpileup --gvcf with the new -g, --gvcf
  option. New options -F to control filter logic and -0 to set missing
  data to REF. Resolved a number of longstanding issues (#296, #361, #401,
  #408, #412).

• norm: Bugfixes (#385,#452,#439), more informative error messages (#364).

• query: %END plus %POS0, %END0 (0-indexed) support - allows easy BED
  format output (#479). %TBCSQ for use with the new csq command. Bugfixes
  (#488,#489).

• plugin: A number of new plugins:

  • GTsubset (thanks to @dlaehnemann)
  • ad-bias
  • af-dist
  • fill-from-fasta
  • fixref
  • guess-ploidy (deprecates vcf2sex plugin)
  • isecGT
  • trio-switch-rate

  and changes to existing plugins:

  • tag2tag: Added gp-to-gt, pl-to-gl and --threshold options and
    bugfixes (#475).
  • ad-bias: New -d option for minimum depth.
  • impute-info: Bugfix (49a9eaf).
  • fill-tags: Added ability to aggregate tags for sample subgroups, thanks
    to @mh11. (#503). HWE tag added as an option.
  • mendelian: Bugfix (#566).

• reheader: allow muiltispace delimiters in --samples option.

• roh: Now possible to process multiple samples at once. This allows
  considerable speedups for files with thousands of samples where the cost of
  HMM is neglibible compared to I/O and decompressing. In order to fit tens of
  thousands samples in memory, a sliding HMM can be used (new --buffer-size
  option). Viterbi training now uses Baum-Welch algorithm, and works much
  better. Support for gVCFs or FORMAT/PL tags. Added -o, output and
  -O, --output-type options to control output of sites or regions
  (compression optional). Many bugs fixed - do not segfault on missing PL
  values anymore, a typo in genetic map calculation resulted in a slowdown and
  incorrect results.

• stats: Bugfixes (16414e6), new options -af-bins and -af-tags to control
  allele frequency binning of output. Per-sample genotype concordance tables
  added (#477).

• view -a, --trim-alt-alleles various bugfixes for missing data and more
  informative errors should now be given on failure to pinpoint problems.

General changes:

• Timestamps are now added to header lines summarising the command (#467).

• Use of the --threads options should be faster across the board thanks to
  changes in HTSlib meaning meaning threads are now shared by the compression
  and decompression calls.

• Changes to genotype filtering with -i, --include and -e, --exclude (#454).

The [foo-1.x].tar.bz2 download is the full source code release. The “Source code” downloads are generated by GitHub and are incomplete as they [don't bundle HTSlib and] are missing some generated files.