MutaGon Calling: Benchmark 4 (call for parGcipaGon) Adam Ewing, - - PowerPoint PPT Presentation

▶
mutagon calling benchmark 4
SMART_READER_LITE
LIVE PREVIEW

MutaGon Calling: Benchmark 4 (call for parGcipaGon) Adam Ewing, - - PowerPoint PPT Presentation

Dec 18, 2022 320 likes •500 views

MutaGon Calling: Benchmark 4 (call for parGcipaGon) Adam Ewing, UCSC TCGA 2 nd Annual ScienGfic Symposium MutaGon Calling Benchmark Process GSC sequences and aligns samples Pair of matched Normal and Tumor BAMs MutaGons

slide-1
SLIDE 1

MutaGon Calling: Benchmark ¡4

(call ¡for parGcipaGon) ¡

Adam Ewing, UCSC TCGA ¡2nd Annual ScienGfic Symposium

slide-2
SLIDE 2

MutaGon Calling Benchmark ¡Process

MutaGons are called

GSC sequences and aligns samples Pair of matched Normal and Tumor BAMs

ParGcipant ¡1 ParGcipant ¡2 ParGcipant ¡3 ParGcipant ¡4

VCFs are collected and compared

slide-3
SLIDE 3

Background / History

MutaGon types: SNV (single nucleoGde variant) INDEL (inserGons and deleGons < 100 bp) ¡ SV (inserGons, deleGons, duplicaGons, inversions > 100 bp) ¡ CNV (copy number variaGon)

History: ¡

  • Benchmark 1: SNVs on six pairs of whole ¡genomes ¡
  • Benchmark 2: SNVs on 14 exomes
  • Benchmark 3: SNVs on 25 exomes with validaGon
  • Benchmark 4: SNVs, INDELs, SVs, and CNVs on whole ¡

genomes ¡from ¡cell lines ¡

slide-4
SLIDE 4

Goals: why do another benchmark?

  • TCGA must ¡measure and set ¡standards for the

accuracy of mutaGon calls

  • Evaluate performance on INDELs, SVs, CNVs
  • This is a controlled experiment:

– Simulate normal contaminaGon

  • Mix tumor and normal cell line data ¡

– Simulate subclonal expansion

  • Mix spiked-­‑in mutaGons
  • Wide ¡parGcipaGon: ¡cell line data ¡is public
slide-5
SLIDE 5

Goals: why do another benchmark?

Andrey Sivachenko, ¡Broad ¡InsGtute

Cancer ¡genomics ¡depends ¡on somaGc ¡mutaGon calls ¡

slide-6
SLIDE 6

Samples for benchmark ¡4: Cell lines

  • HCC1143 / HCC1143 BL
  • HCC1954 / HCC1954 BL
  • Available from ATCC, sequenced at Broad:

– HCC1143 (50x) – HCC1143 BL (60x) – HCC1954 (58x) – HCC1954 BL (71x)

  • All data ¡distributed through CGHub
slide-7
SLIDE 7

Benchmark ¡4: Modeling heterogeneity

  • Three parts to mutaGon calling exercise:
  • HCC1143 (50x) vs. HCC1143 BL ¡(60x)
  • HCC1954 (58x) vs. HCC1954 BL ¡(71x)
  • Simulate normal ¡contaminaGon ¡and subclone expansion for both:

(30x)

  • Total: 28 .bam files, ~4.3 TB
slide-8
SLIDE 8

Public ¡.bams on CGHub

hIp://cghub.ucsc.edu/benchmark_download.html Chris ¡Wilks

slide-9
SLIDE 9

New evaluaGon tools for VCF ¡

  • VCF is a successful standard

– ExisGng VCF tools: e.g. VCFtools, ¡GATK, ¡PyVCF, ¡ etc.

Benchmark 4 is sGmulaGng the creaGon of new tools:

  • ­‑ Bamsurgeon
  • ­‑ VCFcomparator
  • ­‑ Le9Shi9Breakends
  • ­‑ VCF to MAF converter (Thanks to Sage!)
slide-10
SLIDE 10

How to parGcipate

(and/or get ¡more informaGon)

  • Everyone is welcome!! ¡
  • Sign up at poster 64
  • E-­‑mail ewingad@soe.ucsc.edu

Mailing list: tcga-­‑mutaGon@soe.ucsc.edu (contact ¡Chris Wilks: cwilks@soe.ucsc.edu)

slide-11
SLIDE 11

Thanks!

David Haussler ¡(UCSC/CGHub) ¡ Singer Ma (UCSC) ¡ Chris ¡Wilks (UCSC/CGHub) ¡ Mark ¡Diekhans (UCSC/CGHub) ¡ Su Yeon Kim ¡(UC ¡Berkley) ¡ Gaddy Getz (Broad InsGtute) ScoI Carter (Broad InsGtute) Andrey Sivachenko (Broad InsGtute) Mara Rosenberg (Broad InsGtute) UCSC Cancer Genomics ¡Hu (CGHub) ¡ UCSC ReconstrucGon & Cancer groups

slide-12
SLIDE 12
slide-13
SLIDE 13

How? (SNVs)

slide-14
SLIDE 14

Good call (20% ¡alt.)

slide-15
SLIDE 15

How? (SVs)

slide-16
SLIDE 16

Examples: DeleGon (50% ¡MAF)