Hi to the experts,
I am new to RNA-Seq analysis and tried using limma on my data. After voom transformation, I get something like in the image attached. I double checked the metadata used for creating the DGElist and they all seem fine. Has anybody seen a trend like this? I cannot figure out what might be causing this. If you could please explain, would be more than great. I am really stuck in this and want to get out of it so badly... The link to the image is here: https://drive.google.com/open?id=0Bx6HyClNrtOXS1RpMnJYUE9ZZEk
Thanks a lot in advance.
•
written
by
shanika.amarasinghe •
Hi all,
I managed to find out the reason behind this. I have used improperly mapped BAM files in the initial steps. I reckon it's a rookie mistake that no one should do, but could happen if you are as careless as me. So trying to double check the BAM file quality after mapping would save you from lots of heartache.
Cheers!
•
written
by
shanika.amarasinghe •
For future reference I wasn't able to view your image, it wanted me to sign in. Maybe it needs to be shared differently, or maybe it is just me.
•
written
by
jake.hagen •
I'm really sorry about that. Yes there was a permission problem with it. Thank you for pointing it out.
I observed the same mean-variance trend in my data. I checked the BAM file quality and the MAPQ is above 30 for around 95% of the reads. So that shouldn't be my problem. When I filtered for variance > 1 and count > 50, voom gave me a linearly increasing a mean-variance plot. Do you have any suggestions? My pipeline is HISAT2 -> Stringtie -> limma (Ballgown to extract read count from Stringtie generated .gtf files)
Thanks!
Log in to answer this question.
OP, did you ever figure this out? I see something similar if I don't filter out lowly expressed genes but do still see a strong positive relationship after doing so