Ribo-seq vs RNA-seq read count

Question

[ashkan]

Hi Guys,

I have RNA-seq and Ribo-seq data. I have align them to the human genome and got the read count per gene. in my output file the number of read count/gene in RNA-seq is lower than Ribo-seq. my question is that should it be higher in RNA-seq or my results are correct? thanks

[Benn]

How many reads per experiment? I mean how deep did you sequence?

I can imagine that RNA-seq would sequence more nt than Ribo-seq, thus with same number of reads you would have more reads per gene with Ribo-seq. But of course it also depends on how many genes were expressed in you RNA-seq experiment...

[Nicolas Rosewick]

Depends on the sequencing depth I guess. Could you edit your question to add some informations (how did you count the reads, which library prep, sequencing depth, etc...)

[ashkan]

to prepare the library our sequencing facility used this: TruSeq RNA Library Prep Kit v2 to count the number of reads I used the following command:

python -m HTSeq.scripts.count -f -h bam -r pos -m intersection-nonempty -s no -a 10 accepted_hits.bam gencocd de.v19.annotation.gtf > 285_Ctrl_counts.txt

Answer 1

[Devon Ryan]

You're measuring different things with the two datasets. RNAseq is measuring the amount of RNA present. RiboSeq is measuring the amount of rRNA bound RNA present. The latter is showing either active translation or pausing, the former steady state transcription. For a given gene, there's no a priori reason to expect higher lower relative signal in one dataset versus the other.

Of course, if you're just comparing raw numbers without accounting for the number of alignments or anything else then that's a different issue...

[ashkan]

Hi Davon, to analyse Ribo-seq data I filtered out rRNA, tRNA, adaptors and markers and then align them to the reference genome but for the RNA-seq I aligned to the reference genome without any filtering

[Devon Ryan]

You might also need to get rid of snRNAs from the RiboSeq dataset. Are you then seeing huge differences between the resulting percentages being included in the counts or between the raw counts? If the latter, how many alignments are going into htseq-count in each?

Answer 2

[afeizi]

RiboSeq does not measure the amount of rRNA bound RNA present, Ribo-seq measures the ribosome protected fragments (RPFs) of mRNA being translated in a sample. Therefore, reads from Ribos-seq experiment can either represent the real translation of the mRNA or only can be the result of stochastic binding of the ribosome to mRNA due to scanning. It is crucial to define measures such as ORFscores or FLOSS for each ORFs based on Ribo-seq data to distinguish useful Ribo-seq data from noise.

For defining ORFscore, you can read this article from Bazzinni -> https://www.ncbi.nlm.nih.gov/pubmed/24705786

For defining FLOSS, you can read this article from Ingolia -> https://www.ncbi.nlm.nih.gov/pubmed/?term=ribosome+profiling+reveals+pervasive+translation+outside+of+annotated+protein-coding+genes