VarDict Development Log - AstraZeneca-NGS/VarDict GitHub Wiki

11/10/2014

This update contains many improvements as well as bug fixes. Below list some of the updates:

• It now handles BAM files aligned using Novoalign, where it's possible a deletion is immediately followed by an insertion, which haven't been observed in BWA.
• Fixed the negative read count problem for indels, which will result errors in downstream Fisher exact test for somatic mutations or strand bias.
• Improved ability to call complex variants, where multiple deletions and insertions happened in a single read.
• Improved calling tandem insertions (15 bp or longer).
• Adjust allele frequencies for long indels with repeats if no read spanning the indels.
• Changed the deletion quality to use the higher base quality of the flanking two bases.
• In amplicon awareness calling, deletions that extended into another amplicon's primer won't be called as amplicon bias as the other amplicon won't work in theory.
• Raw VarDict will output all variants in a given position, instead of just the most frequent one.
• Change AD field in VCF to be more consistent with GATK.