Peak caller for RIP-seq analysis - required or not?

Question

[priya.dalvi]

Hello everyone! I am a biologist trying to analyse my generated RIP-seq data (+/- treatment, pulldown with histone biding proteins). I have performed genome mapping using HiSat2 and not sure which package to use for Peak calling. I tried Piranha, but the output from it doesn't seem right. I tried to find published work but there are not many people who have performed RIP-seq. Any suggestions regarding the peak caller would be really helpful. Also what would be the appropriate step after peak calling? Just peak annotation and motif analysis? Or no peak calling is needed. In this case how could I proceed further. I would highly appreciate any help in this direction. Thank you very much in advance.

[priya.dalvi]

Thanks for your comment. I tried RIPSeeker, but it a bit complicated for me as a beginner to compute it,. As is it is RNA-IP I assume there are many split reads so I performed Hisat2 mapping. But, I am still unsure of MACS2 after Hisat2

Answer 1

[colindaven]

Not really sure, but I'll be facing this problem soon enough too.

Try MACS2 for peak calling as one of the classics. There are a few bioconductor R packages on RIP-seq. They should give you an overview on further steps.

https://bioconductor.org/packages/release/bioc/html/RIPSeeker.html

Was HiSat2 necessary, i.e. did you see many split reads ? What was your read length? Also, just manually converting your BAMs to bigwig and looking at a few prominent peaks will get you a long way.

[GiV17]

Hi, I used STAR instead Hisat2. Is it Correct?

Answer 2

[xiaocong3333]

Hi, Priya. Have solved the problem? I have RIPseq data to analysis. I tried RIPSkeer and MACS, but the results seem not good. Could you tell me what software you use and the parameters? Thank you!

[GiV17]

RIPSkeer doesen't work in the latest R and Bioconductor Versions. How Can I solve this? Thanks

Answer 3

[xiaocong3333]

I used macs and the parameters are: macs2 callpeak -t ${sample[0]}.sorted.bam -c ${control[0]}.sorted.bam -f BAM -g dm -n ${sample[0]} -B -q 0.01 --nomodel --nolambda --extsize 200 > ${sample[0]}.log

But the results are weird. There are almost 15000 peaks. My data came from fly.

[bhanu.chandra1]

Hi xiaocong3333, have you found a solution? I too have RIP-seq data of the fly model. Looking for a peak caller for motif analysis.