Blocky p300 ChIP-seq reads

Question

[alphaflylizard]

I recently performed a p300 ChIP-seq on two different genotypes.

I aligned my reads using STAR aligner, and made UCSC .bedGraph files that I loaded on IGV viewer. I noticed that the reads for one of my genotypes - genotypeX (red - combined replicates; purple and pink - individual replicates) look very blocky. The reads for the other genotype - genotype Y (black - combined replicates; green, blue - individual replicates) look better!?

Hnrnpa0 is a known target of p300 which is why I chose it as an example.

What could be the reason why the reads look 'blocky'?

How can I troubleshoot this?

Could it be due to duplications or just very bad ChIP?

I attempted to run samtools markdup. I had the following line of commands set up for one my genotype replicates:

Convert my sam files to bam format:

samtools view -b genotypeX_R1.sam > genotypeX_R1.bam

Then I followed the samtools markdup steps as follows:

• samtools sort genotypeX_R1.bam -o sorted_genotypeX_R1.bam
• samtools index sorted_genotypeX_R1.bam
• samtools markdup -r sorted_genotypeX_R1.bam marked_duplicates_genotypeX_R1.bam
• samtools flagstat marked_duplicates_genotypeX_R1.bam

Are these the right set of commands to use to mark up duplicates in my files?

I was introduced to bioinformatics recently and am still struggling to figure out/trouble shoot errors that come up in my analysis workflow. 80% of the time I have no idea how to approach or what I am doing to solve any issues with my data. :(

Thank you so much for your time and help! Appreciate it!

Answer 1

[ATpoint]

If this screenshot is representative for the dataset then it means that the ChIP failed, as there is no peaks and just noise. Call peaks with macs2, a failed ChIP will manifest in very few to no peaks.

[alphaflylizard]

Yes the ChIP is from my data! Would you say the data for genotype Y is okay?

[ATpoint]

No, from this screenshot it all looks like pure noise.

Answer 2

[Ming Tommy Tang]

it is ChIP-seq data, you should use a DNA aligner. STAR is for RNA-seq which aligns the reads to the transcriptome rather than the genome.

[ATpoint]

STAR aligns to the genome, with a GTF to guide splicing if wanted (during indexing). There is nothing intrinsically wrong with this here if the index was built without a GTF, despite bowtie2 or bwa are more typical choices for these data indeed. @OP you can try another aligner such as bowtie2 to be sure though.