ChIP-Seq Top2 in Yeast

From wiki
Jump to: navigation, search


This wiki page combines the two sequencing experiments of ChIP-Seq on yeast samples udring June / July 2017.

Naming conventions

As this was arrived at half way through, not all datasets are named according to this:

CONDITION: YPD (untreated) or ACF (acriflavine treated)
MUTATIONS: WT (Wild type) or ULS1 (uls1 deleted)
SAMPLE TYPE: INP (input) or IP (immunoprecipitation) or broadpeaks28 (peak files and logLR)

Sample quality

In the above previous analyses, we undertook some Duplicate removel as this showed up to be an issue in the FastQC quality files.

However, the effect of this removal was very low. In any case, the MACS2 program takes care of duplicates so samples were only trimmed for adapters and quality.

Alignment to the S288C Reference

Bam quality, no filtering

First Experiment

Second Experiment

Alignment to the new W303 Reference

Bam quality, no filtering

First Experiment

Second Experiment

Bam quality, filtering applied

The bam files were filtered both for mapping quality and low integrity reads in the following manner:

samtools view -b -F 1820 -q 48 <INBAMFILE> -o ${OUTBAMFILE> -@ <NUMTHREADS>

First Experiment

Second Experiment

Bam quality, 2nd filtering applied

This time bam files were filtered to MAPQ 28:

samtools view -b -F 1820 -q 28 <INBAMFILE> -o ${OUTBAMFILE> -@ <NUMTHREADS>

First Experiment

Second Experiment

The MACS/2 program

The main way of running this program is as follows:

macs2 callpeak -B -t $BAM1 -c $BAM2 -f BAMPE --outdir $IPOUTDIR -g 12.07e6 -q 0.01
  • $BAM1 is the name of the immunoprecipitation alignment file, termed the treatment and therefore given with the -t option.
  • $BAM2 is the name of the normal input alignment file, termed the control and therefore given with the -c option.
  • BAMPE is just a setting informing the program that these a pair-end alignment files.
  • the output directory here is given the IP (rather than the INP) directory for the sample.
  • the size of the genome is then given for the -g option.

MACS automatically produces five files as output:

  • NA_peaks.narrowPeak, a 6+4 bedgraph file with two of the four columns being p-values and q-value (false discovery rate). It's the principal peak-calling file.
  • NA_summits.bed, more a summary of the previous file ... recording the max value for each of the peaks called.
  • NA_peaks.xls
  • NA_treat_pileup.bdg, a high resolution 4 column bedgraph file of the basic signal (floating point). Contiguous loci with same signal are grouped together. It is assumed that it refers to the IP signal.
  • NA_control_lambda.bdg, almost the same as the pileup file, but most probably refering to the INP. The meaning of word lambda is unclear here.

(NA is the default prefix, another can be chosen)

Categorisation as peak

It is not totally clear how MACS decides that a peak should be called. This is apparent when the signal appears to rise and fall, yet MACS does not call a peak.

Signal script

The source for this is from the MACS developer and is detailed here.

Differential Peak-calling

Macs2 incorporated differential peak calling somewhat later into its set of tools. The idea is to take two conditions and compare their peaks.

The pipeline to follow is probably Macs2's recommended one at Differential Binding Events