When will this Technology be Viable?

[Dr. Hybe]

It seems that this technology is off to a slow start. Am I correct in saying this or am I missing something?

[nilshomer]

You are missing to which technology you are referring.

[GW_OK]

Given where he's posting I'll bet he means SMRT sequencing...

[GenoMax]

I think the reference is for PacBio, since the post is under that section.

New technologies always get off to a slow start, no surprises there.

We have one in house. I can attest to the fact that it does have some cool mechanical design features (just the sequencer itself). How things will improve (throughput and accuracy) over time would be the real test for the technology. It may never overtake Illumina in terms of sheer raw throughput, but it will serve specific applications.

[westerman]

Since this is posted in the PacBio sub-group I suspect 'Dr. Hybe' is talking about PacBio and its rather unique sequencing. Although it would have been nice to say so explicitly.

A slow start, sure, I would say that. Doesn't mean that PacBio will be a dead horse in the NGS race.

[nilshomer]

I'll take the blame on that one, I didn't check the forum. No comment otherwise.

[Chipper]

Slow start given the hype, but it seems they have made some nice improvments http://seqanswers.com/forums/showthread.php?t=12554

400 bp reads with 97 % accuracy and single pass reads up to 29 kb sounds useful to me, especially when considering the library preps.

[Dr. Hybe]

Sorry - I was referring to PacBio

Can anyone pinpoint specific challenges to increasing throughput and accuracy. i.e. reagents, mechanical issues, or algorithmic/computational limitations?

Dr. H.

[krobison]

Probably the biggest hurdle to PacBio adoption is that they are about $800K each. The second hurdle is the low read quality -- ~85% single pass accuracy (phred 8). Major error mode is insertions in

Improving single-pass accuracy will require better capturing single molecule events, which is quite tricky. I'm sure this will improve

Read length improvements will probably mostly come from engineering polymerases to better withstand the environment and/or additives to the buffer to achieve the same. The chemistry to be released in Q4 was used for the outbreak E.coli sequencing, and enables amazing read lengths.

Higher loading densities of the chips through some sort of ordered scheme would make a big impact (ideally every ZMW would have a single loaded polymerase, but right now it is just Poisson statistics there). The number of reads per SMRTcell is low to be thinking about applying this to really large genome projects. I think they plan to go to even higher ZMW densities in the future, but this will require upgraded optics & that will be a headache to replace/upgrade the existing instrument base.

The strobe sequencing mode looks promising but has had hiccups early on.

Getting more publications out will help, especially some from the beta sites.

[BBoy]

To achieve all that they will need to do something other than optical detection. I am guessing here, but it seems that most the cost and complexity of the instrument comes from the need for precise alignment of the incoming beams to the ZMWs and then detecting the output of the ZMWs. Light is a terrible way to do this, you need to dump a lot of power to get the small signal out, and that probably tends to heat the solution, degrade polymeraze performance, and drive a lot of cost in terms of engineering the beam splitters and keeping them from drifting over time. As best as I can tell, PACB's next gen plans still hinge on optical, and even that has been scaled back, possibly in an attempt to conserve cash.

Disclaimer - unlike krobinson who writes an excellent blog, I know very little about the biology side. I am a semiconductor guy with a personal interest in the technology since I am convinced that sooner or later everyone will have to do some type of electrical detection to meet the stated goals of the $1000/$100 genome. Ion Torrent has the right idea, albeit in a clunky sounding implementation (there are better ways to detect small charges, which would allow 1T cells, smaller beads, and the resulting longer read length). I am surprised how little attention NabSys is getting, since at least from a detection standpoint their approach appears rather robust and scalable. Perhaps other aspects of their technique are inadequate?

[GenoMax]

A company can have the coolest technology around but if they can't translate that into an instrument that can survive being in less than ideal conditions and still produce data reliably then chances of that technology being a commercial success are slim.

Ion Torrent has the right idea, albeit in a clunky sounding implementation (there are better ways to detect small charges, which would allow 1T cells, smaller beads, and the resulting longer read length). I am surprised how little attention NabSys is getting, since at least from a detection standpoint their approach appears rather robust and scalable. Perhaps other aspects of their technique are inadequate?