DSN Normalization vs Ribo-Zero

[clostridium40]

We are looking to do some RNA-seq projects using Illumina for bacterial gene expression, and we are trying to decide between ribo-depleting with RiboZero and using Epicentre's ScriptSeq kits. Or my supervisor is curious about using DSN for ribo-depleting and normalization using the Illumina kits. We are not really looking particularly for rare transcripts, and it seems based on the papers that the standard is Ribo-Zero or some type of ribo-depletion kit, but I haven't seen or heard a lot about DSN treatment. Thus, I was just curious what people have found as the positives and negatives to each of the systems.

Thanks,

Kerry

[Melissa]

You wanna do gene expression studies but considering doing DSN normalization. You're kidding, right? Unless you're interested to prove that gene expression study can be accurate using normalized samples.

[aarodriguezm]

Hi, DSN treatment is only interesting for non-model organisms or without reference in data bases.
My experience with DSN by RNA-Seq is that you will actually only normalized rRNA according to the protocol from Illumina Evrogen. This is because the RNA-Seq protocol does a fragmentation before RT. The original Evrogen protocol for rare transcripts does not fragment the RNA, they use a so call Mint technology to generate full lenght transcripts and normalized them.
If you want gene expression profiling I recomend you to do a very controled depletion and continue with the RNA-Seq procedure.
have fun
A.

[pmiguel]

You wanna do gene expression studies but considering doing DSN normalization. You're kidding, right? Unless you're interested to prove that gene expression study can be accurate using normalized samples.

I know. Sounds completely counter-intuitive!

But at least one publication demonstrated that it can work. I guess what happens is the highly transcribed fractions of the RNA pool swamp the DSN and prevent normalization of the less transcribed fractions. Anyway, empirically, the counts for the less highly transcribed fractions seem unaffected by the normalization.

Sounds like the real reason not to use it for everything is that it is tricky to calibrate to right level of normalization.

--
Phillip

[scbaker]

Despite the 'normalization' in the title, DSN seems to do very little harm in terms of altering the measure expression levels (apart from very highly expressed genes). But it is a pretty tricky protocol. I think the real advantage for DSN is when you have fairly degraded RNA. Any hybridization pull-down method like Ribo-Zero will suffer badly with degraded RNA whereas DSN will handle it just fine.

[clostridium40]

You wanna do gene expression studies but considering doing DSN normalization. You're kidding, right? Unless you're interested to prove that gene expression study can be accurate using normalized samples.

I have spoken to several people that have done DSN normalization for their RNA-seq experiments, and they have all confirmed that the high expressing genes stay that way. So apparently you should check the literature a little more often. I suggest you start here:

Nucleic Acids Res. 2011 Nov 1;39(20):e140. Epub 2011 Aug 31.
Duplex-specific nuclease efficiently removes rRNA for prokaryotic RNA-seq.
Yi H, Cho YJ, Won S, Lee JE, Jin Yu H, Kim S, Schroth GP, Luo S, Chun J.

To everyone else thank you for the useful information, and confirming what I already suspected about DSN normalization.

[mnkyboy]

I have done a lot of both in mammalian systems and RiboZero is the way to go. We got some great results with DSN when it worked but it can be a little sensitive and the normalization can fail easily. There can also be issues with enzyme lots for DSN and the Epicentre kits have been solid with either the non-magnetic and magnetic systems.

We also did RiboZero with highly degraded FFPE samples and that also worked incredibly well and we got very good transcript coverage.