[krobison]

I don't see it as myopic as just guides to understanding what niches the platform will probably fill; I agree that clinical apps (which put a high premium on speed) will dominate. There's also a huge opportunity for smaller genomics studies, given the smaller upfront cost & the fast turnaround.

In the current form, the cost isn't competitive with Illumina or SOLiD if you really want to pile up reads and bases, such as for de novo sequencing, whole genome resequencing or transcriptome profiling. But, I agree that when they succeed at a few rounds of revisions (if I get my act together, tomorrow's blog post will cover this) it really will be serious competition in those areas with Illumina and SOLiD.

[NextGenSeq]

Even at 150MB per hour run that approaches the GAIIX throughput. Admittedly it's more expensive per bp. Also, supposedly the error rate is high and homopolymer stretches are a problem as with the 454.

[Zigster]

let's not be myopic here - this debate is beginning to sound like someone in 1985 comparing an Apple Laserwriter to a full-blown printing press. Yes the printing press has more throughput but that isn't the point. The PGM (or a successor) will find its way to a clinical setting within our lifetime.

[wraithnot]

I hate to disagree, but I believe that quote is from a stock analyst named Jon Groberg : "In a research note analyzing the acquisition, Jon Groberg of Macquarie Research said that even though Ion Torrent has no "meaningful revenues," Macquarie approves of the acquisition "as we think the technology has the potential to be more disruptive than any other technology near nearing the market."
"In addition, Groberg said that Life Tech plans to scale the system up from a current throughput of 2 gigabases per four-hour run to 4 gigabases per four-hour run in the next year by increasing the number of features on the chip from a current level of 7.3 million features to 12.9 million features."

The previous estimates of ~1,000,000 reads with the 316 chip matches what our friendly ion torrent rep told us two weeks ago. The chip itself has far more than 1,000,000 features, but they are currently only getting useful data from about 1,000,000 of those features per run. I assume the stock analyst just read the spec sheet and never talked with anyone who has actual experience with the instrument. But perhaps I'm jumping to the wrong conclusions since I'm not all that fond of stock analysts.

No the specs are


Thus the output is actually higher than the GAIIX

http://www.genomeweb.com/node/947587

[NextGenSeq]

No the specs are

Life Tech plans to scale the system up from a current throughput of 2 gigabases per four-hour run to 4 gigabases per four-hour run in the next year by increasing the number of features on the chip from a current level of 7.3 million features to 12.9 million features.

Thus the output is actually higher than the GAIIX

http://www.genomeweb.com/node/947587

[krobison]

IonTorrent compares very poorly to HiSeq for really high throughput sequencing -- remember the claim is that each run will generate 150Mb (1M reads x ~150nt) of data for $500; HiSeq will generate on the order of (from memory; feel free to dispute with hard data!) 100Gb for $10K -- if you equalize those then Ion Torrent for $10K will generate only 3Gb (20*150Mb) -- about 1/30th the data for the same consumable cost.

Of course, I've left out equipment amortization from that. If you really use the sequencer flat out, HiSeq should still win (even if Ion comes in at $50K) -- and Ion Torrent will require hourly attention to care-and-feed (the videos I've seen do NOT look automation friendly -- someone snapping a new cartridge in each time), meaning higher labor costs.

On the other hand, to rapidly turn around small batches (such as amplicons), Ion Torrent will probably win -- it will do that run in an hour vs. around a week for HiSeq. Now, the appropriate comparator is really 454 and even with 454 Jr the Ion Torrent looks good -- a lot of data cheaply, if really long reads aren't critical for you.

Ion Torrent claimed that next year they would have a chip with about 4X the density. If priced the same (unlikely, but possible) then it's stlll not the price-performance of HiSeq when generating huge amounts of data from a given sample or set of samples with little time pressure.

[NextGenSeq]

The HiSeq is backordered for so long maybe people should just cancel their order and wait for this?

[greigite]

The projected price of the Ion Torrent used to be 50K- now it is "less than 100K" in the new press releases...wait till we see the reagent costs!

[janejane]

Will the current NGS products soon be replaced or are the new instruments just additive to the market?

[Zigster]

i posted some of my thoughts about the future of Ion Torrent, with an emphasis on clinical applications here:

http://jermdemo.blogspot.com/2010/08/my-thoughts-on-acquisition-of-ion.html

[NextGenSeq]

Competition would have brought down prices. Too bad that LIFE is buying everything now. They and Illumina can basically price gouge.

[Zigster]

The 454 story is more convoluted. Rothberg had already left Curagen by 2007 to purse other interests. Roche was only a minority investor in the beginning. Then Curagen, having no idea how to run a sequencer company and figuring there was more promise in their drug pipeline (that eventually went nowhere), basically asked Roche what they'd like to pay for the remainder of 454, which turned out to be a paltry $152M.

All the investors who were into CRGN for 454 sold out immediately, and the stock plummeted.

[kmcarr]

I don't think it is that shocking. This is similar to what Jonathan Rothberg (Ion Torrent founder & CEO) did when he founded 454. He developed the technology to the point it was ready to go into commercial production and sold out to a company with the resources make that happen.

[KevinLam]

I had a shock as well. totally didn't see it coming though i should have..
Ion Torrent seems to be in direct competition with their SOLID PI system.

[Zigster]

that is unreal
right out of the gate