Directional RNA-seq: Illumina Tru-seq versus dUTP based method

[jazz]

I guess the title says it all here. I am looking for an RNA-seq protocol which preserves strand-specificity, and came across this paper.

Comprehensive comparative analysis of strand-specific RNA sequencing methods.
Joshua Z Levin,Moran Yassour, Xian Adiconis, Chad Nusbaum, Dawn Anne Thompson, Nir Friedman, Andreas Gnirke & Aviv Regev

http://www.nature.com/nmeth/journal/v7/n9/full/nmeth.1491.html

The authors have a done a comprehensive comparison study for various methods, and concluded that the dUTP based method (using dUTP for second strand synthesis, followed by UDG digestion) is the best among 7 others.

Since then, Illumina has come up with their Truseq RNA-seq kit.

I was wondering if anyone here has an experience to share about these 2 methods. Which one is better?

Thanks

[mnkyboy]

Do you want to have paired end data? Do you need to index?

We have been running the old Illumina supported directional and have had great results for whole transcriptome data. However we cannot get paired ends or multiplex this way. I tried to modify this for TruSeq and lets just say I failed miserably which they were not surprised about.

I am actually going to start working on the dUTP protocol soon and it will be interesting how it turns out.

[jazz]

Yes, I failed to mention that I needed to multiplex my samples and have paired-end reads.

[protist]

We have been using a modified version of the Parmuchuk method (Nucleic Acids Res. 2009 Oct;37(18):e123) since 2009 with very few problems. For multiplexed samples we use 'home-brew' in-adapter indexing. I have yet to use TruSeq adapters - maybe if they brought out an oligo-only kit as they did for the original adapters we might switch...

[treebeard]

follow-up: strand-specific RNA-seq

Jazz - we worked the Parkhomchuk et al. approach directly into the flow of the TruSeq mRNA-Seq kit; details are here (http://openwetware.org/wiki/Directional-RNAseq_Prep). The reported page numbers are correct for the version 1 TruSeq kit, but page numbers changed in the v2 manual (easy enough to find). We use the Illumina index adapters for multiplex sequencing (typically 6-plex), and the process works well with paired end sequencing. We suspect that there must be a small amount of carry-over of dTTP in this method, as only 95% of reads map to the expected strand (and we don't believe that the remaining 5% is due entirely to the presence of antisense RNA).

[protist]

Hi treebeard, Have you tried including actinomycin D in your second strand synthesis.

[pmiguel]

We suspect that there must be a small amount of carry-over of dTTP in this method, as only 95% of reads map to the expected strand (and we don't believe that the remaining 5% is due entirely to the presence of antisense RNA).

Any reason not to use AmpPure beads after first strand synthesis? Might be more effective at eliminating the dTTP than an ethanol precipitation.

--
Phillip

[ECO]

In my experience, quite a bit of dNTP/NTP make it through an AMPure.

[protist]

For the past ~2 yrs all of our RNAseq experiments have been strand specific (dUTP method) - we have consistently used g50 columns for 1st strand clean-uo and rarely see problems with the resultant data.

[perrinwang]

In my experience, quite a bit of dNTP/NTP make it through an AMPure.

Hmm... that is interesting to know... also troublesome for us. I am wondering how much dNTP is retained through Ampure purification, do you have a good guess at percentage?

Here is our protocol: http://dx.plos.org/10.1371/journal.pone.0026426, i think we would have the issue if dNTP is carried over.

[treebeard]

Hey all - couple comments:

1. Protist asked "Have you tried including actinomycin D in your second strand synthesis." We use the NEB second strand kit without modification, and it lacks AcD. Do you see a difference in libraries produced with AcD?

2. Eliminating unincorporated dNTPs. This is the reason we used ammonium rather than sodium acetate for preciptation - it is more effective at removing nucleotides. For complete removal, I think that protist has the best suggestion (e.g., G50 columns).

[protist]

Protist asked "Have you tried including actinomycin D in your second strand synthesis." We use the NEB second strand kit without modification, and it lacks AcD. Do you see a difference in libraries produced with AcD?

When I did the initial library testing we didn't see a noticeable difference and the Parkhomchuk NAR paper I adapted the method from stated "In our hands, the presence of actinomycin D in the reaction did not significantly influence the level of antisense transcription". We have not routinely included actinomycin D in our library preps.

[jazz]

While we are discussing the usability of different reagents in preparing libraries, can anyone help me with my issue with Ampure beads. I just ordered them and wanted to test them for size retention using a 100bp ladder. I tried 3 different concentrations- 0.5:1, 1:1 and 1.5:1 for beads to DNA. All three cases, there was no loss of any DNA whatsoever. My understanding is that lower concentration of beads should cause selective retention of large sized DNA (and hence loss of smaller bands). I incubated with the beads for 15 minutes. Is that too much? Can anyone share their experiences/suggestions?

[protist]

pbluescript posted a very nice image showing the effect of altering the ratio of AMPure beads/sample in a Primer dimer problem thread - http://seqanswers.com/forums/showthread.php?t=10817. I have reattached the informative image pbluescript added illustrating the effect of altering the ratio of AMPure beads/sample.

[frozenlyse]

Protist asked "Have you tried including actinomycin D in your second strand synthesis." We use the NEB second strand kit without modification, and it lacks AcD. Do you see a difference in libraries produced with AcD?

When I did the initial library testing we didn't see a noticeable difference and the Parkhomchuk NAR paper I adapted the method from stated "In our hands, the presence of actinomycin D in the reaction did not significantly influence the level of antisense transcription". We have not routinely included actinomycin D in our library preps.

From my results, Superscript III doesn't need ActD to eliminate spurious antisense strand generation, whereas a lot of other RT enzymes do.