Normalisation of small non coding RNA (miRNA and tRNA fragments)

[Lili_FCN]

Hello all,


I have been sequencing small non coding RNA using hiseq 100pb single paired

I then used mirdeep2 to analyse the miRNA portion
I blasted the sequences against the appropriate tRNA database to identifying tRNA fragments in my samples.

And here comes the question:
I first got my miRNA data and used DEseq2 to normalise it and highlight differential expression. But this kind of normalisation does not consider the size of each library/replicate when normalising, which make the comparison between miRNA and tRNAfragment portion not possible.

Is there a way to normalise both miRNA and tRNAfragment dataset for them to be comparable ? Would DEseq2 be OK ?

RPKM seems to be a way to normalise tRNAfragment data (doi:10.1186/s12915-014-0078-0 )

Thanks

[dpryan]

In general, using library size to normalize produces worse results. If you're worried that tRNA contamination is suppressing your read counts in one group and producing spurious findings, then perhaps incorporating the tRNA counts into the initial count matrix and then removing those lines after library normalization would work for you.