Can ncRNAs be found from all RNA-seq data?

Question

[fernardo]

Hi everyone,

There are tools (miRDeep2, deepBase, ... etc) for finding and predicting known and novel non-coding RNAs(ncRNA) from RNA-seq data.

The question is, should the RNA-seq library be prepared specifically for ncRNA (lncRNA, miRNA, ...etc) or every RNA-seq dataset which are used for mRNA expression analysis can be also used to find ncRNA through the tools mentioned?

Thanks in advance

[David Langenberger]

If you want to use one of the tools you mentioned, you need a small RNA-Seq library preparation. In RNA-Seq datasets you might find some short reads (between 18 and 30nt in length), but these are outliers, due to the protocol. In small RNA-Seq experiments, molecules between 18nt and 30nt (or 50nt) are selected and sequences. These short reads are needed to use e.g. miRDeep.

Note: The short length of the reads from small RNA-Seq are not because of a fragmentation! They occur in that length within the cell. In RNA-Seq you do a fragmentation step and in small RNA-Seq you don't.

[fernardo]

Thanks. This is also informative. Means according the the protocol exist so far, they can't be detected at the same time.

Answer 1

[Devon Ryan]

Don't polyA enrich if you want to look at ncRNAs (some will be polyadenylated, others not). Note that you can't look at miRNAs and lincRNAs at the same time, they require different library prep.

[fernardo]

Thanks. The total RNA dataset means polyA + non-polyA? I mean if it is not polyA enriched means that dataset can be used for both mRNA and ncRNA detection?

[Devon Ryan]

Correct, instead of polyA enrichment one does ribo-depletion. You can then detect mRNA and ncRNA.

[David Langenberger]

One add here: You can detect ncRNAs with RNA-Seq and ribo-depletion, but not by using miRDeep. The tool is not programmed for such experiments. At least not the version I'm aware of.

[Edalat]

hi Devon I have some human non coding RNA-seq data. For getting diff-exp of miRNA,I should trim length between 18 to 30 befor starting? why Avg. Sequence length is 51,can I use these data for getting diff-exp of lncRNA too?

[Devon Ryan]

Don't worry about trimming to a particular length. If you trim off of adapters then that will largely take care of itself. Assuming you really sequenced mostly miRNA, then the remaining sequence is adapter and junk, which will get trimmed. Of course if the average length is still 51 after trimming then most likely you have mostly non-miRNA.

[Edalat]

Thank you for your attention. I have several problems with noncoding RNA-seq data, library of some of them is ncRNA-Seq and its other is miRNA-Seq,what is difference? Second, when I use fastqc, it shows that the adapter is trimmed, but length is 51.why?!

[Devon Ryan]

You'd have to ask whoever created the library for details of what they might mean by ncRNAseq vs. miRNAseq. My presumption is that miRNAseq is a standard small RNAseq kit, whereas ncRNAseq is either total RNA (i.e., Ribo-depleted) or ribo and polyA depleted, possibly with some additional size selection.

FastQC doesn't perform trimming. If it reports that everything is length 51 then you haven't done any trimming (or you used the wrong adapter).

[Edalat]

If ncRNA-Seq library means total RNA,for getting diff-exp of miRNA should I trim length before starting small RNA-seq analysis?

is it possible fastqc give wrong report about Existence or type of adapter?

[Devon Ryan]

Just trim adapters, it's rarely worthwhile to trim to a specific length. FastQC will only be wrong if adapters it has never heard of are being used. The odds of that are low.

[Edalat]

you know I compare the adapter that fastqc suggest with that Experimenter said for one experiment,they are quite different and I am extremely confused. Do you have any suggestions?

[Devon Ryan]

What'd the experimenter expect them to be and what did FastQC report? Perhaps you got the wrong samples (it's happened to me before).

[Edalat]

experimenter said: ADAPTOR3=AGATCGGAAGAGCACACGTCTGAACTCCAGTCAC -ADAPTOR5=GTTCAGAGTTCTACAGTCCGACGATC

FastQC report: illumina universal Adapter

This time more carefully, I see that fastQC reported "illumina universal Adapter" but I use "TruSeq Universal Adapter",are they different?

I use "TruSeq Universal Adapter" because in "illumina adapter seq pdf" and the site" https://github.com/csf-ngs/fastqc/blob/master/Contaminants/contaminant_list.txt " one universal Adapter exist that it is "TruSeq Universal Adapter"

[Devon Ryan]

"Adapter 5" is a sequencing primer, adapter 3 is used by a bunch of things in illumina sequencing. Don't trust what the experimenter said, he/she doesn't understand what's being asked. Just trim the reads with "Trim Galore!", which will pick up whatever the proper adapter is. Having said that, if FastQC didn't show up much then it isn't there and there's essentially no miRNA that was sequenced (regardless of what the experimenter is expecting).

[Edalat]

Really?! Thank you for helping and awareness.

Another question, if after trimming true adapter, Avg. The sequence length is not about 22 for example be 30 or 40, something is wrong?

is it possible all of samples of one experiment need trim adapter except one or two?

[Devon Ryan]

Maybe you have a bunch of pre-miRNAs? Just align them and see where they go. It's possible for a couple samples to not have any adapter read through (e.g., if they have few RNAs below the read length).

[Edalat]

No I don`t think.if it is not problem for you I will say result.

[Arindam Ghosh]

What are the different types of transcripts we should restrict ourselves to while analyzing rRNA depleted samples? As mentioned, we should not look for miRNA in such samples.