Eye of the next-gen storm?

[pmiguel]

Anyone else getting that "eye the storm" feeling about the current state of next gen sequencing technologies? Stuff working pretty much to spec. No way to completely adjust to the continuing revisions to the technologies, but to the extent that one becomes inured to constant change -- well, the pace is endurable. The big 3, while not at all releasing their death grips on each others throats, currently are mainly spamming us with various "mini" versions of the instruments we already have. An almost comedic interlude?

Nevertheless the big third gen sequencer releases seem imminent. Pacific Biosciences instruments, I have heard, are being "sold". Not delivered yet, for the most part, but sold. And one is led to believe that some contest winners should get their "Ion Torrent" instruments as some point in the not too distant future. Roche threatens "1K" technology, AB has a major instrument upgrade (v4hq) that will cost ~$90K per instrument -- but these are outside the current 3-6 month horizon of predictability that reigns in the next gen world.

Eye of the next gen storm?

--
Phillip

[GW_OK]

I think we've merely hit another leveling off of technology, much like Sanger sequencing before all this next gen stuff. There only so much you can do before you need to change the paradigm. From what I've heard lately about the Ion Torrent, and what I saw of the PacBio at Marco Island, we're still a year or so out from the earliest real deployment of 3rd gen tech. Eye of the storm? No. Another technological plateau? Yes.

[pmiguel]

I think we've merely hit another leveling off of technology, much like Sanger sequencing before all this next gen stuff. There only so much you can do before you need to change the paradigm. From what I've heard lately about the Ion Torrent, and what I saw of the PacBio at Marco Island, we're still a year or so out from the earliest real deployment of 3rd gen tech. Eye of the storm? No. Another technological plateau? Yes.

Yeah, but I think we are talking about the same thing. The "storm" here being the paradigm change and the "eye" being the plateau.

Maybe lab-side the 2nd to 3rd gen switch won't even be that bad. I personally found the switch from Sanger to 2nd gen required a vast amount of protocol retooling. Specifically switching from bacterially-mediated cloning methods to amplicon and bead work required different skill sets.

If 3rd gen just means no more bead work, that would be one headache gone. But who am I kidding, there is always another headache in the wings ready to replace the one you have...

How about this: The promise of a new technology is generally easier to appreciate at a distance than its pitfalls. I'm not saying that in the Luddite sense. (Although one could make that case as well.) Rather, until you are down in the trenches with the new technology it is difficult to guess what the problems in getting it to work will be. More just a restatement of "the devil is in the details".

--
Phillip

[drio]

There is an interesting interview to Mark Stevenson from Life Tech about this. You can download the pdf of the magazine for free online.

[GW_OK]

On further reflection I wonder if the 3rd gen will actually be much of a "storm". I agree that Sanger to 2nd gen was a huge, major shift, but now people are realistically talking about whole genomes or other previously giant projects in a more "easily do-able" way. With 3rd gen tech, we'll still have amplicons I think (gotta have someplace to put a sequencing primer), and will still have beads (at least for the Ion Torrent). But the overall thought processes of the projects won't change, I think. We'll just have more data to slog through and it'll cost less to get it (which means you can run more samples through, which means more data, LOL).

Maybe there'll be a technology not requiring anything more than DNA shearing, if that.

tl;dr 3rd gen isn't going to be as big a deal as 2nd gen was.

[krobison]

The next wave of sequencers will increase the storm intensity, if only because the low price point for entry will mean many more labs getting into the game. That will mean a lot of projects -- some well done, some awful. Microarrays went through a similar shift.

Standard sample prep for PacBio appears to be a single ligation reaction after shearing. I haven't heard what it is for the Life Tech (ex-Visigen) technology. Oxford Nanopore or some of the other nanopore techs might start to approach shear-and-go.

[pmiguel]

Maybe there'll be a technology not requiring anything more than DNA shearing, if that.

There are a series of technologies that could lead to this -- I think of them as "4th generation". Those not requiring an enzyme at all. Nanopore or automated atomic force microscopy would be examples. That is, something that "reads" the DNA (or RNA) directly.

Hey, since we are dreaming anyway, why limit it to polynucleotides? A general purpose bio-polymer reader to read proteins, polynucleotides and poly-saccharides would be welcome, I'm sure.

[GW_OK]

I recall a guy at AGBT talking about reading protein sequence using a PacBio with ribosome and mRNA instead of polymerase and DNA.

Anyway, I think we're witnessing the death of the do-all sequencing technology. Some technologies are better than others at different things and I think the (big 3) companies are starting to figure that out and market themselves accordingly.

[pmiguel]

The next wave of sequencers will increase the storm intensity, if only because the low price point for entry will mean many more labs getting into the game. That will mean a lot of projects -- some well done, some awful. Microarrays went through a similar shift.

And who feels the storm. Moore's law in the past outstripped sequence capacity increases so much that throwing more computational power at most problems was generally the most economical way to solve a problem.

We are past that point now, with sequence costs declining at a much higher rate (and acceleration) than the decline in cost per calculation. So the "storm" is moving inland. Of course there is still slack in the system. Even simple stuff like file compression and binary data file formats are only just starting to be used generally for sequence data.

But it is not clear to me when sequencing hits its "red brick wall". I think the way to look at is: how many eyes are on the problem? Vast resources go into the research and development driving Moore's law. A tiny percentage of those resources are spent on the research and development driving down the cost of sequencing. That leaves a lot of head room to deploy more resources.

--
Phillip

[pmiguel]

I recall a guy at AGBT talking about reading protein sequence using a PacBio with ribosome and mRNA instead of polymerase and DNA.

Wait. You are talking reverse translation here? I'm not following the mechanism. An mRNA is not degraded by the process of translation, so it isn't like you can force the reaction backwards and end up synthesizing an RNA encoding a given protein.


--
Phillip

[pmiguel]

There is an interesting interview to Mark Stevenson from Life Tech about this. You can download the pdf of the magazine for free online.

link?

--
Phillip

[GW_OK]

Wait. You are talking reverse translation here? I'm not following the mechanism. An mRNA is not degraded by the process of translation, so it isn't like you can force the reaction backwards and end up synthesizing an RNA encoding a given protein.


--
Phillip

Looking back through the notes and such, it was Joe Puglisi from Stanford looking at tRNAs transitioning through the ribosomal complex during translation using the PacBio.

Article:

http://www.ncbi.nlm.nih.gov/pubmed/20393556

So you could theoretically sequence a protein by watching it get made, I guess.

[steven]

Looking back through the notes and such, it was Joe Puglisi from Stanford looking at tRNAs transitioning through the ribosomal complex during translation using the PacBio.

Article:

http://www.ncbi.nlm.nih.gov/pubmed/20393556

So you could theoretically sequence a protein by watching it get made, I guess.

.. which means that you actually sequence the coding section of the corresponding mRNA and not the to-be-made protein, i guess. thanks for the link.

[drio]

link?

--
Phillip

http://www.qmags.com/qmct.asp?q=114&qt=1049580&u=/d.asp?pub=BIW&upid=15381&s=ML&fl=others/BIW/BIW_20100501_May_2010.pdf

[james hadfield]

If there is going to be a new storm I suspect it will come in the form of an instrument that can take millions of proteins and 'sequence' them. How is a big unknown but this would revolutionise things in the Proteomics field.

I wonder if anyone is anchoring proteins to a slide and digesting away one amino acid at a time, in real-time, monitoring at the single molecule level? Do amino acids work with a system like Oxford Nanopore?