• What makes somatic SNV detection difficult?

• CaVEMan and some other calling tools

• Statistical approaches to calling somatic SNVs

• How well should you expect a tool to perform?

Somatic SNV callers typically set the expected mutation rate to be around 5 mutations per megabase, i.e. a total of 15,000 mutations across the genome.

Source: ICGC Data Portal

• Low cellularity (tumour DNA content)

• Intra-tumour heterogeneity in which multiple tumour cell populations (subclones) exist

• Aneuploidy

• Unbalanced structural variation (deletions, duplications, etc.)

• Matched normal contaminated with cancer DNA

  • adjacent normal tissue may contain residual disease or early tumour-initiating somatic mutations

  • circulating tumour DNA in blood normals

• Sequencing errors

• Alignment artefacts

Mwenifumbo & Marra, Nat Rev Genet. 2013

In this example the tumour was sequenced to an average depth of 50.

• Is this sufficient?

• Consider the 50 observations of our tumour which carries a mutation at this base

Tumour cellularity

• In fact the 'tumour' sample has some normal contamination

• 40% of our reads could easily be from the normal sample

Tumour heterogeneity