Over-represented sequences

[starbug16]

Hi,

I have one lane (~360 million reads) of Illumina transcriptome data which I am trying to assemble. Having only a tiny server, I am attempting to reduce this dataset with digital normalization. However, before doing this, I have run fastqc which flags up some over-represented sequences. When I BLAST these, I get matches to my organism's 16S, 18S and 28S rRNA genes.

As a bit of a beginner, I was wondering whether this was normal and whether I need to worry or do anything about this?

Any help or advice very much appreciated!!

[dpryan]

That's completely normal, don't give it a second thought.

[mastal]

It's quite normal, especially if the library prep was done using total RNA, and not using any methods to reduce or remove ribosomal RNA.

[starbug16]

Phew, thank you for the responses. Mind much at rest and now onwards to diginorm-ing!

[yueluo]

If you're not interested in ribosomal genes, you might as well remove all those reads.

[Marianna85]

Hi everybody,

actually I had the same problem (that is not really a problem but something that is quite normal) and I want to ask you an opinion that it might be helpful also for starbug16.

I did several libraries by using different kits and I had different percentages of reads mapping to rRNAs, ranging from 1 to 30%.
I think that the problem arises when you have to compare samples having very different percentages of rRNAs. I mean: if I have to find DE between 2 samples having respectively 15% and 30% of rRNA reads, I have the impression that final results are biased by the very different percentages of rRNAs (which compete for the sequencing and affect the number of mRNA sequences). Did you understand my point?
In this case, do you think that a normalization procedure (like TMM or DESeq) will minimize this issue?

Thank you to anybody who will tell his opinion!

Marianna

[yueluo]

I usually remove reads that map to rRNA(and/or other sources of contamination) , then proceed with mapping/DE-analysis .

[Marianna85]

Hi Yueluo,
so you think it is not a problem if your samples have really different percentages of rRNAs??

Marianna