Chip-Seq Peak Calling Using Macs In Galaxy- Overlapping Peaks For Replicates

Question

[anon111]

I have two replicates from one cell type (chip-seq data). I aligned each one in galaxy using bwa. I then did peak calling using MACS for each one separately because I was unable to successfully merge the BAM files and then peak call them together in one BAM file.

Now I have the peak models for each replicate and want to note the number of peaks observed. How do I this? Also, how do I overlap and compare the peaks of the two replicates so I can note the overall number of peaks?

Answer 1

[David]

You can try to merge your results at the level of MACS by following the instructions here

1) align with BWA output as SAM.

2) Run macs2 on both separately (you obtain bedGraph)

3) Transform to bigWig with bedTool

4) Check your replicate are correlated and if yes merge them with macs2

Answer 2

[spacemorrissey]

MACS should output a bed file of peaks. You can then use the intersect tool under Operate on Genomic Intervals to find out how many peaks the replicates have in common. If you want the union of the peak calls, concatenate the two bed files, then use the merge tool under Operate on Genomic Intervals.

Also, You should be able be able to use the merge Bam files tool under sam tools to pool both replicates. If not you can concatenate the sam files and then convert them to bam files.

You may want to look more closely at the reproducibility of the replicates by using IDR which unfortunately is not in Galaxy yet.

[Ian]

If they are biological replicates I would perform two different MACS analyses as suggested here and not pool the reads. If they are technical replicates (of the same sample across different lanes) I would convert the BAM file to BED format and concatenate the BED files. bedtools has a tool to do the BAM to BED conversion.