1000 Genomes SV detection Boston College Chip Stewart 24 November 2008

Spanner: RP approach Paired-end SV breakpoint detection 1.Detect clusters of fragments spanning breakpoints 2.Classify clusters into SV types 3.Estimate CN from read counts in candidate region 24 November 2008Spanner / BC2 LM ~ LF – L dup high coverage inversion LM ~ +L inv LM ~ -L inv normal coverage ends flipped L inv tandem duplication L dup LM ~ LF + L del low coverage deletion LM LF L del LM DNA REFERENCE Pattern: PE RD

Spanner … RD approach Read Depth Copy Number Variation (CNV) detection 1.Count uniquely aligned reads in windows (1kb) across chromosomes. 2.Correct for micro-repeat artifact with an “alignabilty” metric to estimate expected number of uniquely aligned reads in each window. Local copy number is based on the likelihood that the estimated count of reads scaled by copy number will fluctuate (Poisson) to the observed count. 3.Identify CN breakpoints (this is still in the works) Alignability: Define “alignabilty” for a given chromosome position as: This becomes tractable when considering a random sample of possible reads within a window of positions. 24 November 2008Spanner / BC3

Alignabilty 24 November 2008Spanner / BC4 The count of unique aligned read (up to 4 mismatches) in non-overlapping windows of 1kb. compared with the expected count of reads based on A(p). Coverage profiles from NA12878 for a 1.2MB region of chromosome 4. Reads/1kb Chromosome 4 position [Mb] Position [Mb] expected Reads/1kb A(p) has a 90% correlation with observed read coverage variance. Reads/1kb expected Reads/1kb

SV event display matlab tool 24 November 2008Spanner / BC5 chromosome overview fragment lengths read depth event track 300bp ALU deletion in chromosome 1 of NA12878

Tandem duplication event 24 November 2008Spanner / BC6 NA12878 chromosome 1

“Complex” SV 24 November 2008Spanner / BC7 NA12878 chromosome 8

Trio CNV events detected 24 November 2008Spanner / BC8 sampledeletionstandem duplicationscomplex NA NA NA