Genomic Contamination with Clontech PolyA selection?

[travelk]

Hi Everyone,

I am working with single cell RNA-Seq and am currently analyzing my data. I've noticed that I often get a lot of reads aligning to the UTR but not at all to the exons. When I look closer, I see that there is often a poly A region here.

So, my question: Poly A tail selection using the Clontech SMARTer technology is supposed to only target mRNA (and lncRNA). However, is it possible it also sometimes catches genomic DNA? Or will the fact that it's double stranded prevent the Clontech adapters from binding to the region? Anybody else have problems with genomic contamination? Or perhaps you have another theory why I'm seeing a lot of specific alignment to to multiple areas with genomic polyA regions?

Thanks for your insight!

[morgane51]

Hey I don't know about this kit but this happened to me with a Truseq kit.
Anyway one should avoid to have genomic DNA in the starting sample...
We noticed it before loading on the Hiseq because the genomic DNA hadn't been fragmented like the RNA. Was obvious the Bioanalyzer, we got excessively long fragments. Did you check this out?
Did you do a DNase step before prepararing the libraries?
If you still have RNA (I guess not...) you could make an intron test.
what do you mean with " I'm seeing a lot of specific alignment to to multiple areas with genomic polyA regions?"
Cheers

[travelk]

Thanks for the response!

I don't do a DNase step because I use the Fluidigm C1 system and it's done within the chip in the machine. The tech sheet specifically states: "Ribosomal*RNA*(rRNA)*removal*or*DNase*treatment*of*RNA*samples*is*not*required*for*these*kits.These*kits*selectively*and*efficiently*amplify*polyA+*RNA*regardless*of*the*presence*of*rRNA*or*genomic*DNA."

We do run all our samples on a Fragment Analyzer but the upper marker is set at 6000 bp and none of our samples get up that high so it didn't set off any flags. At the same time, when I look at the output, it doesn't show anything higher than that so I don't know if the genomic DNA would be missed because it is off the scale. I will look further into that. Thanks for the suggestion.

In regards to the alignment, I use StringTie to look for potential new lncRNA and of course I do a differential expression analysis (with DESeq) between two different cell types. Occasionally, I'll find a DE target that statistically looks great, but when I look closer on IGV, there aren't any reads that align to the exons, there are only reads that align to the UTR. At first I thought maybe it was an interesting regulatory RNA element but I noticed the same pattern in numerous genes. When I looked closer at the actual sequence, I often found a polyA region. This led me to be suspicious that during our polyA selection, it might be picking up genomic DNA. The same with a lot of the new transcripts found by StringTie. They often end with a poly A region in the genome sequence (although I know this can be possible with non-coding RNA transcripts).

Or, if it's not genomic DNA, perhaps the data is being biased by polyA sequences found in UTRs of mRNA (although I would have thought that this would cut off the end of the UTR rather than the translated exons).

I'm not sure if that made it any clearer or more confusing...