ChIP-Seq Analysis allele specific way?

Question

[Susmita Mandal]

Hello everyone,

I am about to do ChIP-Seq analysis allele specific way. I have not done ChiP-Seq analysis practically, but I have some idea about how to do it. The problem is I'm not getting any answers for doing ChIP Seq allele specific way. I have done allele specific RNA Seq analysis. I have the two parental genomes as well. From what I have read so far, after aligning the ChIP-Seq, we do peak calling and it gives us a bed file. So

1. Is it same as RNA-Seq? Should I align the ChIP reads to both the parental genomes?
2. After peak calling, should I intersect the bed file to my vcf file like in RNA-Seq or there is another way of doing ChIP-seq allele specfic way?
3. Or it is not possible altogether and I'm barking the wrong tree?

Any help is appreciated.

Thanks, Susmita

[liang795]

Hello, If I used the separate bam file (1 and 2) to call peaks using macs2, do you know how I should set up the effective genome size for it? Is it the same as the effective genome size for the bam file without separating 1 and 2? Thank you so much!

[Ankit]

Hi,

If you have the solution for peak calling for allele specific chip seq data, please let me know.

I also ran SNPsplit and I have two separate bam files (allele 1 and allele2) for both Chip n input. I was thinking how to call the peaks for chip sample using input.

I thought for possibilities for peak calling using MACS/MACS2, please correct me:

1. Chip-allele1 and input-allele1 and Chip-allele2 and input-allele2.

2. Chip-allele1 and merged (bam or bed) input-allele1+input-allele2. Similarly for Chip-allele2 and merged (bam or bed) input-allele1+input-allele2.

3. Rather than calling peaks from allele-specific data, use the regular chipseq peaks (bulk) and find enrichment of allele specific reads under regular peaks. This will quantify and will give count matrix and can help for predicting monoalleleic binding between two allele (something like differential binding).

Please suggest.

Thanks

[Susmita Mandal]

Merge those two bam files, run the peak calling with the input. Then split the peak calling file using intersectBed toold from bedtools, basically intersect the peak calling file with your vcf file.

Also can you tell me how you did SNPsplit?

[Ankit]

Hi. Thanks for the response. So I have to do something like I mentioned in point 2. I am not sure if I have to again split peaks by overlapping with vcf. Because Genomes are already splitted SNPsplit

For SNPsplit, I created a N-masked Genome using SNP information between two mouse strains. I used a custom python script to incorporate Ns at the Known SNP sites into reference Genome. You can try doing same with a tool. Thanks

[liang795]

Hello, do you think if we have to specify different effective genome size when using macs2 to call peaks for file 1 and file 2 separated from bam file? It looks allele-specific alignment rate is much higher in one genotype than another, but if specify the same genome size, would that be a problem? Thank you!

Answer 1

[Devon Ryan]

1. Yes, it's the same. Yes you should align to both (possibly then using SNPsplit).
2. It's best to call peaks on the merged sample and then quantify according to parent of origin.
3. It's totally possible.

[Susmita Mandal]

Thank you for the response. I have few questions though. How to do SNPsplit? On both the differnet alignment files or should I merge it. and regarding calling peaks, what is the merged sample and how to quantify according to parent of origin?

[Devon Ryan]

Read the SNPsplit documentation and everything will be clear.

[Susmita Mandal]

So far I understood that 1. Align the sample to both the parental genomes. 2. Merge the SAM/BAM file. 3. Use SNPsplit to get tags for allele specific. 4. Run the BAM file for peak calling

Correct me if I'm wrong. And also how to quantify?