Metagenomic eukaryote/prokaryote RNASeq: ratio of counts to be expected?

[BaCh]

Dear all,

I'm going to do RNASeq on a eukaryote including the prokaryotes which populate it (skin, digestive tract etc.). I am wondering about the numbers of transcripts one has to expect coming from the different organisms when going into an experiment like this, I could not find any numbers anywhere.

There are a couple of factors which may play a role, like what exact eukaryotic host one looks at etc.pp. E.g., for a typical human, it has been reported that the number of cells from colonising organisms (mostly bacteria) outnumbers the number of cells from the human host by a factor of 10 to 1. Then again, the human genome is approximately 500 to 1000 times larger (leading to more transcription opportunities), cells are larger (more space for transcripts to persist), etc.pp

So, as a thought experiment: if we'd take a complete human, flash freeze it, grind, homogenise, extract RNA from the whole mixture and sequence that ... what would be the ratio of bacterial vs. human transcripts? Any ideas?

Best,
B.

PS: just in case it's not obvious: I am of course not doing this with human or any other higher eukaryote

[ZWB]

Hello B,

I've been doing some bacterial transcriptome analysis from human clinical samples and found that the portion of bacterial RNA is very small. Typically we find ~0.01% of the reads map to bacteria in the blood of septic patients. This is further complicated by the fact that >90% of the bacterial reads are rRNA. If you are clever about how you collect and process your samples you can increase this percentage dramatically however. For example if you take the luminal contents of the small intestine rather than the entire intestine, or if you take a skin swab and not a chunk of skin tissue you will get much more bacterial transcripts in your mix. In addition, if you know the bacterium that you are looking for we have developed some methods that separate known bacterial transcripts from host transcripts. If you are interested... http://www.ncbi.nlm.nih.gov/pubmed/23535274

Zach

[BaCh]

Typically we find ~0.01% of the reads map to bacteria in the blood of septic patients.

Ah, but indeed one expects to have very little bacterial content in blood as this tends to be a rather unhealthy situation for patients. However - to stay with the thought experiment from above - if you crushed a complete patient you'd have ~1 to 1.5 kg of bacterial material from guts/intestines and maybe a couple of grams more from skin surface. That's the kind of situation I am wondering about the ratio.

This is further complicated by the fact that >90% of the bacterial reads are rRNA.

Good point, I forgot to mention that we will do a rRNA depletion step. I know, I know, we won't catch everything.

Best,
B.