Interprete gene body coverage plot

Question

[wax4001]

I am wondering what could cause the jump in the plot ?

The plot is generated using Qualimap. BOTH the jump sample shows weird expression data that I am trying to understand why and maybe troubleshoot. I am begin to believe that the sample have DNA contamination but I am wondering how to explain this jump.

Answer 1

[Jianyu]

Usually, this kind of jump is due to an over-represented sequence. In your situation, maybe one or two over-represented sequences located in the 60%~70% position of the transcript caused this jump. If you have some programming skills, you can try to plot a heatmap showing the coverage of each transcript to find out which transcript has this kind of sequence.

[wax4001]

I do find some over-represented sequence. However, this is geneBody coverage plot. if a transcript is over-represented, why would it cause a jump at the position and that is like 10 bp ?

[Jianyu]

If an over-represented sequence is located in the middle of a transcript, then you will see a jump in the middle.

The length of transcripts has a large variation, some transcripts only have less than 100bp. So 10bp over-represented sequence occupied 10% region on these transcripts

[wax4001]

so you are saying only a few gene's few sequence got overrepresented in the case and causing this overall jump ? I have check the alignment distribution across the genome which looks normal. DO you think removing the overrepresented sequence would be a way to correct the data ?

[i.sudbery]

One possibility is high expression of a small RNA that is embedded in a longer sequence. For example, we still this sort of a pattern being caused by snoRNAs when we do iCLIP or NET-seq analysis.

[Jianyu]

Yes, I have seen some data having this problem, but there could be some other reason I don't know.

I don't know how you find the alignment distribution normal, whether to remove the overrepresented sequence should be done after you have precisely known why this jump happens.

I still recommend you to see the coverage per gene first to make decision