• Functional impact of indels in cancer

• Short exercise: exploring TGF-beta receptor 2 indels in COSMIC

• Indel calling tools

• How well do somatic indel callers perform?

• Some characteristics of indels called in HCC1143

• Range from 1 to 10,000 bases but here we are mostly considering small indels of 1 - 50 bp

• Lower frequency than SNPs except near highly repetitive regions, including homopolymers and microsatellites

• 1000 Genomes Project loss of function indel variants in each individual [Nature 2010]

  • 340 - 400 premature stop codons, split site disruptions and frame shifts

  • 250 - 300 genes affected

• Frameshift mutations have an impact in several diseases (cystic fibrosis, HIV, Crohn's disease, Tay-Sachs)

• Colorectal cancer

  • 15% of colorectal tumours characterized by microsatellite instability (MSI), caused by defective mismatch repair

  • DNA slippage within coding sequences induces frameshift mutations resulting in truncated, functionally inactive proteins (TGFBR2 and BAX frequently targeted genes)

• COSMIC database of somatic mutations in cancer has several examples of frameshift mutations in tumour suppressors, such as TP53, PTEN, BRCA1/2 and TGFBR2

• In-depth analyses in large cohort studies typically focus on substitutions or copy number aberrations/rearrangements

  • Indels often just catalogued

• Recent Sanger Institute paper on 560 breast cancers searched for novel indel drivers in non-coding regions with significant recurrence (functional regulatory elements) [Nik-Zainal et al., Nature 2016]

  • Previously looked at association of indels with mutational signatures [Nik-Zainal et al., Cell 2012]

• Application of 1000 Genomes Project data to cancer genomics showed that genes linked with cancer showed stronger selection against indels [Science 2013]

• Search for TGFBR2 in the COSMIC website

• Explore the breakdown of mutation types for samples with TGFBR2 mutations (Distribution tab)

  • What proportion of samples have frameshift indels?

  • What size are most of these indels?

• Look at the location of the indels within the gene (Gene View tab)

  • What is immediately striking about this?

  • Zoom in on the most frequently observed indel and turn on the DNA sequence display

  • What is the sequence context at this indel?

• Click on the c.374delA deletion to access details of the cancers in which this mutation was observed

  • Which cancer type is this mutation most commonly seen in?

• Several tools call both SNVs and indels particularly those tools that perform local reassembly or realignment around likely indel sites

  • VarScan2
  • Strelka
  • MuTect2
  • VarDict

• Pindel is the somatic indel caller used in the Sanger CGP pipeline

  • Uses a pattern growth algorithm for unmapped paired end reads anchored to mapped mate

  • Can identity short and medium sized indels up to 10kb

  • Modified version used in CGP pipeline that takes advantage of additional alignment information available for longer reads (>100 bases) and can make use of split read alignments

• Many indels exist in long homopolymers and short sequence repeats (di-nucleotides, tri-nucleotides, etc.)

• Not easy to distinguish between true variants caused by replication slippage and sequencing errors

  • Strelka filters out indels within repeats above certain length

• Homopolymer A/T indels are major source of low quality indel calls [Fang et al., Genome Med. 2014]

• Benchmark exercises show that indels are harder to call accurately that SNVs

  • ICGC medulloblastoma benchmark study [Alioto et al., Nat Commun. 2015]

Total of 208 somatic insertions and 564 somatic deletions called by Pindel.

272 hompolymer A/T indels (35% of all indels), almost all of which are 1-bp insertions or deletions.