My question is more or less the title of the post:
Total RNA was isolated at various time points from bacterial samples that were infected by a phage. I carried out rRNA depletion and generated sequencing libraries using NEBNext kits.
As the time goes on, I see more and more reads mapping to the phage genome as the concentration of phage transcripts in the cell increases. This of course means there are fewer reads mapping to the bacterial genome and I'm concerned that it could appear as if genes are being downregulated when their transcripts are actually just being "crowded out" by the phage transcripts.
Does it make sense then to split the data once I have my count matrix into bacterial counts and phage counts, and then do normalization and differential expression analysis separately? Or am I thinking about this all wrong?
Any input would be very much appreciated!
What is the aim of this experiment? That would be good to know. Nothing should be getting crowded out since you would be counting the reads at the same time (I assume you included phage sequence as an extra chromosome in alignments). Is the phage sequence completely new i.e. there are no endogenous pre-existing sequences with similarity in bacteria?
The aim is to look at the host response to the phage infection in a wild type strain, as well as in a mutant strain.
I think a "bonus" aspect is to have a look at how the transcription of phage genes progresses over the time course, as well as to see if the phage transcripts differ in the wild type vs. mutant.
I could very well be overthinking this, but this is my first run at doing RNAseq and I want to ensure I'm getting it right.
EDIT:
Sorry, didn't see your last bit - yes, included the phage sequence as an extra chromosome in alignments and the phage sequence is completely new - there are no similar sequences in the bacterial genome.
I wonder if you need to analyze the host sequences independently of phage (increasing amounts of phage is physiological response). Is this a lytic phage or it simply takes over all cellular machinery from the bacterium. Would be very difficult to separate the hijacking of cellular machinery from actual host response (perhaps I am overthinking this).
It is a lytic phage... I actually hadn't considered how that would affect things. I'm not an expert on the biology of the system, just doing the library prep, sequencing, and data analysis.
Also curious. I've read several host-virus papers that do DGE and I don't see any of them account for the changing ratio of host to viral DNA present in the cell. Biologically, there's a limit of RNA polymerases so by mass action, you shouldn't need to account for the transcriptome shift (read: captured decreases in RNA abundance are true decreases in abundance) but just because I haven't seen anyone do it doesn't necessarily mean that's still the correct way of doing it...
I was trying to look at other papers that have done this and I also didn't see anyone accounting for this. I had considered your point about a limit of RNA polymerases, so perhaps it is indeed the case that I shouldn't need to account for it just based on the biology.
I would hazard a guess that there is an excess of reagents compared to the sample (you would not see the increasing amount of phage otherwise). There may indeed be some limiting factor in terms of how well the conversion to cDNA works but perhaps that limit is applicable for all samples being handled (as long as they are processed at the same time).