Hi,
We have conducted a bacterial RNAseq using rRNA depletion on a NextSeq. Upon analysis, it seems some samples still have a significant amount of reads (5-40%) mapping to rRNA genes. How does this affect downstream analysis? I was planning on analysing using Rockhopper, but don't know if rRNA is disregarded for normalisation, or how differential expression is calculated based on this. Any suggestions?
Regards.
Hello,
I had simillar problems with different rates of rRNA depletion from different samples of eukaryote (Arabidopsis). The rate of depletion depends greatly on the hybridization efficiency of LNA probes, which can be affected by insufficient control of hybridization temperatures and heating/cooling rates.
I solved it by filtering out rRNA reads prior to aligning to whole genome. Simply prepare Bowtie2 reference file with rRNA sequences and run Bowtie2 on it, allowing for no mismatches. Save the unaligned reads and use for standard alignment to full reference.
This step should work extremally fast, since the reference file is small.
If you perform it correctly, all normalizatin methods should work just fine.
Best regards,
Maciej
Dear Wiep,
as far I know (I'm working with RNAseq from bacteria and yeast since two years ago), around 1% of rRNA is still detected with RNAseq after rRNA depletion. If you detect more, depletion has not worked at all.
Rockhopper include also rRNA in normalization and differential expression call. Unfortunately, Rockhopper reports the expression level for each transcript (RPKM) by correcting it with the upper quartile of gene expression instead by the total number of reads, which does not allow for comparison within samples with different success in rRNA depletion.
You can try to use RPKM counts table and then perform differential expression call by using R.
Or use RobiNA, which also allows you to determine counts table for each transcript (correct by total number of reads and gene length), which allows you to perform downstream analysis with other software or even with in-home R.
hope this helps you,
Beatriz Sabater-Muñoz
TCD
Hi Beatriz,
Thanks for your comments, that is what I thought. The service provides claims that anything below 50% rRNA mapped reads is still considered depleted, as the ribosomal RNA bands are virtually non-detectable when they run the sample on a chip.
I have found somebody willing to run a RPKM calculation for me with a normalization over total reads that exclude rRNA. So I will work with this.
Regards,
Wiep Klaas
Hello,
I had simillar problems with different rates of rRNA depletion from different samples of eukaryote (Arabidopsis). The rate of depletion depends greatly on the hybridization efficiency of LNA probes, which can be affected by insufficient control of hybridization temperatures and heating/cooling rates.
I solved it by filtering out rRNA reads prior to aligning to whole genome. Simply prepare Bowtie2 reference file with rRNA sequences and run Bowtie2 on it, allowing for no mismatches. Save the unaligned reads and use for standard alignment to full reference.
This step should work extremally fast, since the reference file is small.
If you perform it correctly, all normalizatin methods should work just fine.
Best regards,
Maciej
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Molecular Biological Techniques