3' Bias

[Carly Graham]

I am currently analyzing an RNAseq dataset with 18 different samples, all of which were prepared using the same techniques. I ran fastqc on all of the samples followed by STAR genome alignment and Qualimap. The MultiQC report indicated that one of the samples has a very low 5'-3' bias of 0.22. This same sample also has a high duplication rate and low % aligned to the genome. I have attached the output table and coverage profile image.

Because all of these samples were prepared using the same approach at the same time, I am assuming this is not a prep difference issue. Could this be a result of low RNA input quality or quantity? What is the best way to account for this high 3' bias? I would prefer to not have to remove the whole sample.

Thanks in advance!

Carly

[Konstantin Okonechnikov]

Hi! 

Taking into account that all other samples show clear good quality from the same run, this issue is indeed related to some preparation or low input, not sequencing procedure. I would also recommend excluding this sample since it will introduce various differences due to 3' bias effect and will be most likely an outlier.  

Hope that helps,

   Konstantin

--
You received this message because you are subscribed to the Google Groups "QualiMap" group.
To unsubscribe from this group and stop receiving emails from it, send an email to qualimap+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/qualimap/5b651bc1-3e42-419c-85e6-94c112e4bc3en%40googlegroups.com.

[Carly Graham]

Hey Konstantin,

Thanks so much for the response! 

In my reading this week I have come across a couple of tools that use models to try and correct for biases such as this. Some of the programs are BCseq and Mix2, do you have any experience with these? 

Carly

[Konstantin Okonechnikov]

Hi Carly,

nope, did not work with such tools... Had another experience:  combination of 2 sample cohorts where one of them was with 5' loss. Solution was batch effect adjustment via Combat. But for such a technique of course it's assumed that at least 2 samples represent each batch cohort. 

Best regards,

   Konstantin

To view this discussion on the web visit https://groups.google.com/d/msgid/qualimap/e03f74f5-8d3d-4433-9a07-73ea15e35bfen%40googlegroups.com.