ISO-seq library preparation, is it size selection RNA classes or RNA fragments?

[Fad2012]

Dear all

I am preparing to do PacBio ISO-Seq sequencing for total RNA of HeLa cells infected with 7.5Kb virus, I am interested in the Viral transcripts. My question is about library preparation and divided into two parts. Part 1 of the question: after isolating the polyadenylated transcripts, which step comes first, cDNA synthesis, or size selection? the second part of the question is about the size selection process, will the RNA go through a shearing process and then size selection, or the size will be selected based on RNA classes?

As I am interested only in the viral transcripts which is about 7.5Kb, thus size selection based on the classes will give me a stronger signal. While in the case of size selection based on fragments of sheared RNA will lower the signal, please correct me if I am wrong as I am not sure how to deal with the coverage if RNA has to be fragmented before selecting the size

Many thanks

[rhall]

The full protocol can be found here.
Short answer, size selection comes after cDNA synthesis, and in order to maintain full length transcripts no shearing occurs.

[Fad2012]

Thank you very much rhall.

This is great, so the size selection is based on RNA classes. The protocol you mentioned is showing the size selection for maximum 6kb, is it also possible to do it for ~7.5kb so I can retrieve the whole viral transcripts?

[rhall]

It isn't a hard maximum, just that the yield for fragments larger than 6kb will be low. Catching the 7.5kb viral transcript will depend on how abundant it is. You could sequence without size selection, but the 7.5kb will be such a low percentage of all the transcripts (and shorter sequences preferentially load) that you would have to sequence a lot to catch it. If you size select at >6kb hopefully the relative abundance of the 7.5kb transcript is such that you can retrieve it within a reasonable amount of sequencing.

[Fad2012]

Hi rhall

Thank you very much for your informative response. I found from a webinar on 11th of February that PacBio can reach to 10kb (photo). But I can't find this in a published protocol. I was wondering also if we can add a cDNA normalisation for the protocol, which will also enrich for the rare transcripts, is it possible in the context of PacBio ISO-seq lib-prep? In other words, after selecting the size, is it possible to do a step of DSN normalisation ( duplex-specific nucluase digestion) to select the rare transcripts of 7.5kb?

Many thanks

[rhall]

From an isoseq expert, re. DSN:

Entirely depends on the amount of material they get out…they need to make sure they still get enough material to be able to go into library prep afterwards. The biggest issue will be the amplification of the longer transcripts. IsoSeq itself wouldn’t be affected I believe, as this protocol doesn’t change the insert itself, and you’ll still have diversity and relative abundance in the fraction that is amplified.

Just to reiterate, though, we don’t officially recommend normalization, but rather size selection of your 1st round of amplification cDNA first, then optimizing your large scale amplification followed by size selection after library prep.

[Fad2012]

Enhance the chance

My concern is that we are taking a risk here, as we are relying on the leaking-through 7.5kb molecules from the 3kb-6kb fraction. I was wondering, by using the protocol you suggested, is it possible to reach the distribution of 5-10kb molecules as it appears in the attached photo. If not, is it possible to use specific viral primer to amplify the 3kb-6kb fraction, as the 5’ of the virus is more important to us. The reverse primer could remain the Clontech one and the forward one will be a specific viral primer that primes at position 1.5Kb of the viral genome (This will generate a product of 6kb). By doing this we enrich for viral molecules, and at the same time shorten the molecule to decrease the leak-through risk, thus increase the chance for the viral molecules to be better represented in the 3kb-6kb distribution?

Many thanks

[Fad2012]

My concern is that we are taking a risk here, as we are relying on the leaking-through 7.5kb molecules from the >6Kb molecules. I was wondering, by using the protocol you suggested, is it possible to reach the distribution of 5-10kb molecules as it appears in the attached photo. If not, is it possible to use specific viral primer to amplify the 3kb-6kb fraction, as the 5’ of the virus is more important to us. The reverse primer could remain the Clontech one and the forward one will be a specific viral primer that primes at position 1.6Kb of the viral genome (This will generate a product of 6kb). By doing this we enrich for viral molecules, and at the same time shorten the molecule to decrease the leak-through risk, thus increase the chance for the viral molecules to be better represented in the SMRTbell library.



Many thanks

[rhall]

You can certainly use different Blue Pippin cutoffs, but as you can be seen from that photo, although the cut is 4.5-10kb (on the Blue Pippin) the peek is as at ~6kb. Ideally for the best 7.5kb enrichment you would want to shift this distribution up a little, by using a higher cutoff window.
Priming from viral sequence will certainly give you a better chance of capturing the transcript if it is in low abundance, the only caveat is that will break the default Iso-Seq analysis workflow, but it shouldn't be too difficult to modify the analysis.