Hey Sachin,

Strand specific libraries allow you to identify the coding DNA strand for the transcript you sequence, while the strand information is lost in what you call regular library protocols. Neither polyA selection nor ribo-minus gives you information or limits to whether or not you will create a stranded library. Both are simply mechanisms to enrich the amount of interesting RNA (especially getting rid of rRNA, which makes up most of the total RNA). PolyA-based protocols select polyA mRNA only (be aware that some encoding RNAs might not have polyA-tails) and if you are interested in any other RNA that is not polyadenylated (microRNA, snRNA), you should use a ribo-minus approach. For your purpose polyA might be sufficient. Stranded libraries should be preferred, because they give a more complete view of the transcriptome and might significantly alter differential expression results. With stranded libraries antisense transripts can be identified as such, whereas with unstranded libraries you will have no idea as to whether these sequences originated from which strand and whether it was the same.

The following literature might help you to decide for an appropriate method:

A comparison between polA and ribominus:

Cui et al. 2010: A comparison between ribo-minus RNA-sequencing and polyA-selected RNA-sequencing

A comprehensive overview of strand specific techniques and benefits:

Mills et al. 2013: Strand-Specific RNA-Seq Provides Greater Resolution of Transcriptome Profiling

Good Luck!

Cheers,

Leon

Hi Leon, thanks for your detailed reply. If I have understood correctly, so a polyA library can also give stranded library, am I right? I see an Illumina kit which makes Directional RNA (polyA-selected) library, which is the same as stranded, isn't it? So should we go for stranded polyA library instead of regular (unstranded) polyA library for my purpose, please let me know? Thanks.

Another confusing term: directional and non-directional libraries. are the same as stranded and unstranded? If they are two different things, what is the difference?

Hi Sachin,

the stranded (=directional) kit is the standard kit now (at least in our sequencing facility), so I think you should go for that. As Leon pointed out, there can be antisense transcripts, which are transcribed at the same locus as some genes but at the opposite strand (search for FLC1 in Arabidopsis for example). Therefore, if you don't have the information about the strand, you'd incorrectly add the count for the antisense transcript to the count for the sense transcript. Just be aware that counting the reads might be a bit more complicated if you use paired-end as well.

Hope that helps,

Nicole

Hi Nicole and Leon, this means I should go for stranded polyA library, am I right? BTW, my samples are rice tissues. Thanks.

If you are not interested in non-coding RNAs then poly-A enrichment is what you want, yes. For my experiments it always gave better results but that might be species specific.

Cheers,

Nicole

But even with poly-A enrichment, we should go for stranded poly-A instead of unstranded one, is that what you mean? Sorry for asking again. Thanks.

Exactly! :)

Right! :)

Hi Nicole and Leon,

Following my question from the last week, if I do polyA strand-specific, will I need to do 100 bp paired-end sequencing or just 50 bp single-end will do? We initially thought of doing 50 SE for non-stranded, will 50bp SE work for strand-specific too? Please let me know. Thanks

Hi Sachin,

you can get strand specific reads for paired-end and single-end reads. Also, the length doesn't matter. 

Which combination is best for your experiment depends on a few things: 

50 bp single-end would definitely be fine. The 100 bp reads can add specificity, if you have a lot of similar genes (e.g. gene families). You will have more near identical 50 bp than 100 bp stretches between different genes. On the other hand, if your reference genome is not the same species as your reads the differences between the species could prevent the 100 bp reads to map at all.

I think that paired-end reads only really help if you want to assemble the transcriptome. You do get twice as many reads but this is misleading because the mates actually come from the same fragment. 

So, I don't think paired-end is necessary in your case. If you have to pay more for the 100 bp reads than I'm pretty sure that you will be fine with 50 bp unless rice has an awful lot of very similar gene families. I haven't worked with rice yet so don't know about that...

Does this make sense?

Cheers,

Nicole