Are labs moving quickly to so-called "third-generation" sequencers with accordingly long read lengths? Or is this only used by a handful of people for the time being, for various reasons (too new, too expensive, etc.)? If your lab will soon use this technology, which platform do you intend to use and for what kinds of experiments?
We are using the MinION and are expecting the PromethION in a few months. Since we work on human genetics, the MinION hasn't been of enormous use, yet, because the throughput isn't sufficient for genome sequencing and enrichment for long reads is not really easy. But we're working on that, too.
We are also using MinION essentially for some targeted sequencing of custom long-range librairies or to solve complex badly annotated genomic regions. Its use is still quite marginal for us but it seems like something really promising.
We will be getting a Promethion soon, but that's not really because we've found Nanopore reads particularly useful so far. They are much harder to deal with than PacBio. I don't think anyone here is really clamoring for it... it's probably more to hedge out bets and stay cutting-edge than anything. That said, the PacBio RSII data has been indispensable in producing high-quality assemblies of repetitive genomes, but we are throughput-limited.
I would rather see a high-quality 2x500bp platform from Illumina, myself, compared to a Promethion.
1) PacBio has very consistent quality; my experience with Nanopore is that it is highly variable.
2) PacBio has fairly consistent yield - not as good as Illumina, but still predictable. My experience with Nanopore is that yield is wildly variable. This is fine when you have plenty of DNA, but absolutely unacceptable for situations in which DNA is limited or difficult to obtain.
3) PacBio has a fairly random error profile; Nanopore errors tend to be highly biased and correlated, making it much more difficult to correct.
4) PacBio reads are loops. You can sequence shorter fragments with multiple passes, increasing the overall quality. For example, I'm currently working with fungal full-length 18S data. These are... 1600bp or so? They get lots of passes, maybe 20, and can yield reads with 99.5% accuracy.
Bear in mind, my observations are not particularly recent. Most are from over a year ago.
Thanks for clarifying, definitely point 4 is a big advantage for PacBio, but obvious trade-off versus read length. For Nanopore sequencing the past year has brought quite some improvements in accuracy (90-95%) and throughput (3-7Gbase). ICYMI, there is recent wgs data from NA12878 on MinION (multiple flow cells), see https://github.com/nanopore-wgs-consortium/NA12878
Has the Sequel delivered results as promised?
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
Are there applications (that are commonly applicable) where the current generation sequencers need to be replaced with the next thing?
Every new generation of sequencers induces a "must keep up with the Joneses" tendency that forces people to consider a purchase even though it may not be needed.
Genome assembly and structural variant identification hugely benefit from longer reads.
We do have PacBio, 10x genomics, BioNano genomics now. In fact no sequencing technology can do what BioNano genomics can. We can get better results by using BioNano and sequencing together.
Do you know of any
third"next"-generation technologies that are significantly better than what we currently have?Ultimately the focus of "
thirdnext generation" technologies is diagnostic sequencing. We may get side benefits for research along the way.Well Pacbio is 'third generation sequencing', 10x genomics is clever second generation sequencing and bionano genomics isn't sequencing at all. I think the most complete genome assemblies have been made integrating BioNano genomics, long read sequencing and high accuracy short read sequencing so they're pretty complementary.
Oh, forgot to say earlier that both PacBio and Nanopore allow detection of nucleotide modifications. Oxford Nanopore is currently testing direct RNA-seq (without cDNA-conversion) which will really open pandora's box with regard to RNA-modifications. Obviously long read transcriptome sequencing has huge implications (nice papers from PacBio IsoSeq for genome annotation.)
That boils down to how do define 'better'. Single nucleotide accuracy? No. Longer reads, faster identification of pathogens, high quality genome assembly: pretty much yes.
I was not able to see the PDF @Alex had linked yesterday but did so now.
Let us set aside the generation part. We really are not seeing truly "next" generation techs (which should be the 4th?) since PacBio, 10x, Oxford have all been around for a couple plus years at this point. (Perhaps we may see something out of the left field at AGBT in February)
While companies claim to do "X" have you seen many publications describing these modifications (IsoSeq is more straight forward and clearly the best application in that space). My understanding is that these kinds of experiments are extremely finicky and hard to do/reproduce.
You are right, the literature with regard to nucleotide modifications is limited. But I think returning to this thread in 6-12 months will provide entire different insights.