v0.4.0

Added

• Added new command line option --global_align to both command line tasks
  vcf_eval and make_truth_vcf. This is meant for small (eg covid) genomes,
  where both truth and reference are one sequence only. It is less efficient
  but more accurate.

• Added command line options that only apply when --global_align is used:
  --global_align_min_coord, --global_align_max_coord,
  --sanitise_truth_gaps, --use_non_acgt,

• If using --global_align, writes an MSA of rev and truth sequences, and
  a FASTA of the truth sequence, but with santised gap lengths.

• New tag in VCF file VCF_QRY_VARIANT, which is the variant with respect to
  the query sequence, as opposed to the ref sequence.

Fixed

• various edge cases caught when there are indels and/or Ns in sequences

• edge case caused by indels when making VCF with --global align.

Version 0.3.1

Same as v0.3.0, except this has the correct version in the code, so --version does the right thing

Version 0.3.0

• add option to not use maxmatch option with nucmer, for big genomes
• expose minimap2 and nucmer threads options
• improved handling of indels, in particular fixes a bug where FP indels could be called as TP, and also now uses current latest mummer version 4.0.0rc1 for better variant calls for recall.
• small bug fixes: handle whitespace in fasta header lines, non-ACGT characters, and lowercase characters

Version 0.2.0

Backwards incompatible output compared to v0.1.0. Changes:

• added --filter_pass option to up front filter records in input VCF
• only reports one set of precision and recall, not filtered and unfiltered
• algorithm change to fix problem with clustered SNPs and indels. Was incorrectly calling TPs as FPs in cases where indels caused ref/probe alleles to not be aligned in the right place. Solved by applying all filtered variants to the probe flanks.

Version 0.1.0

Initial release