tRNA expression analysis from RNA-Seq

Question

[Nicolas Rosewick]

Hi,

How can I assess the expression of tRNAs from RNA-Seq. Is it possible to do it with the "classic" align on the genome and count gene per feature strategy ? or a separate alignment step is required ( such as align only on tRNA sequences). Is smallRNA-Seq better for this purpose as tRNAs are not very long ?

Thanks in advance for your adviced and ideas

[Manvendra Singh]

Yes,

get tRNA sequences , index them, map your reads with bowtie.

I do it quite often to analyze Ribo-seq data, where I remove the reads which map on tRNA or rRNA sequences.

hth

[Michele Busby]

Just to expand on what Manvendra said, tRNAs aren't normally poly adenylated (unless they're marked for degeneration) so you need to look at the protocol that was used to generate the data and see how rRNA was depleted. For many protocols this is done with an oligo that sticks to poly A tails. So look for a protocol that use whole RNA and something like RiboZero.

Answer 1

[mark.ziemann]

IF you have RNA-seq data already, I would suggest to use a standard RNA-seq pipeline (Tophat+cufflinks or STAR+featureCounts) to align all RNA-seq reads to the genome and summarise the expression counts. Then after edgeR or DESeq analysis you can extract the tRNA accession numbers from the spreadsheet. tRNAs are annotated in the ENSEMBL GTF file. You might even make a custom tRNA gene set and use GSEA (preranked).