Cleaning ChIP-seq data

ChIP-seq with Bioconductor in R

Peter Humburg

Statistician, Macquarie University

Common Problems

Incorrectly mapped reads may produce false peaks.

• Genomic repeats.

Common Problems

Incorrectly mapped reads may produce false peaks.

• Genomic repeats.

• Incomplete reference sequence.

Common Problems

Incorrectly mapped reads may produce false peaks.

• Genomic repeats.

• Incomplete reference sequence.

• Low complexity regions.

Amplification Bias

• DNA fragments extracted from cells are copied multiple times prior to sequencing.
• Not all fragments produce the same number of copies.
• Multiple copies of the same fragment may be sequenced.
• A single DNA fragment may inflate coverage and lead to incorrect peak calls.

Quality Control Reports

library(ChIPQC)
qc_report <- ChIPQC(experiment="sample_info.csv", annotation="hg19")
ChIPQCreport(qc_report)

Preparing input files

SampleID	Factor	Condition	Tissue	Treatment	bamReads	Peaks	PeakCaller
S1	AR	primary	primary prostate tumor	gleason score: 3+4=7	S1.bam	S1.bed	macs
S2	AR	primary	primary prostate tumor	gleason score: 3+4=7	S2.bam	S2.bed	macs
...	...	...	...	...	...	...	...

Cleaning the Data

• Remove duplicate reads.
• Remove reads with multiple hits.
• Remove reads with low mapping quality.
• Remove peaks in blacklisted regions.

Cleaning the Data

• Remove duplicate reads.
• Remove reads with multiple hits.
• Remove reads with low mapping quality.
• Remove peaks in blacklisted regions.

Cleaning the Data

• Remove duplicate reads.
• Remove reads with multiple hits.
• Remove reads with low mapping quality.
• Remove peaks in blacklisted regions.
  • Blacklisted regions are available from the ENCODE project.

Let's practice!

ChIP-seq with Bioconductor in R