Entering edit mode
5 months ago
bioinfo1994
▴
20
hello, I downloaded an RNA-seq dataset and aim to determine if it contains evidence of a viral infection. The presence of a specific virus in the sample was confirmed by lab tests mentioned in the associated paper. However, I couldn't detect the virus in the RNA-seq data. I used FastQ Screen and found no reads, and ran Kraken2, which identified only 30 viral reads out of 500,000 unmapped reads.
Any idea why this might be happening? Could someone recommend a more effective tool for this analysis? Thank you!
I was involved in the development of a pseudoalignment-based translated search virus detection workflow earlier this year: https://www.biorxiv.org/content/10.1101/2023.12.11.571168v2.full
It works well in detecting viral infection if you are interested in giving it a try.
However, more generally in answer to your question, you'll need to give more details about your virus and your RNA-seq data. Viruses are very diverse, e.g. some are DNA, some are RNA, +/- strand, etc. Sequencing assays are also fairly diverse: e.g. some use poly-dT priming, some use random oligomer priming, etc. Do you expect, based on the biology of your virus and the chemistry of your sequencing assay, that you should actually be able to detect the virus? Furthermore, it is unsurprising to have low number of viral reads even with infection (think about all the highly expressed mRNAs, rRNAs, etc. that would be present in your dataset and that could be hogging up the overwhelming majority of your reads).
30/500k would be 60 counts per million! And assuming 95% mapping rate, that'd be 3 counts per million mapped reads and given the small genome size that could be significant (maybe Im thinking about it wrong?)
How many reads would you get if you were to map with another uninfected set? Or to a comparable, but divergent virus?
Also are the viral reads well mapped, or is it just partial reads that align?