Skip to content

Variant classification in paired analysis

Zhongwu Lai edited this page Jun 9, 2015 · 1 revision

In paired analysis, VarDict will classify each variant into the following types that are propagated into STATUS info tag after var2vcf_paired.pl script.

When both samples have coverage for the variant:

  1. Germline: detected in germline sample (pass all quality parameters)
  2. StrongSomatic: detected in tumor sample only
  3. LikelySomatic: the variant has at least one read support OR allele frequency < 5% (defined by –V option with default 0.05)
  4. StrongLOH: detected in germline sample only, opposite of StrongSomaitc
  5. LikelyLOH: detected in germline but either lost in tumor OR 20-80% in germline, but increased to 1-opt_V (95%).
  6. AFDiff: detected in tumor (pass quality parameters) and present in germline but didn’t pass quality parameters.

When only one sample has coverage for the variant:

  1. SampleSpecific: detected in tumor sample, but no coverage in germline sample (it’s more technical than biological, as it’s unlikely a tumor sample can gain a piece of sequence in reference that germline sample lacks).
  2. Deletion: detected in germline sample, but no coverage in tumor sample

These are only rough classification. You need to examine the p-value (after testsomatic.R script) to determine whether or not it's significant.

Clone this wiki locally