[victorsor]

I also don't understand the comparison between the Proton and the Miseq? Isn't Proton at the HiSeq 1000 price segment? Have the prices changed, I though after server, extra stuff the Proton/Chef/support ended up at 300k?? I when we bought our MiSeq the PGM was offered to us at $150k when it was supposly $50k.

Does anybody know an updated price table for 2013?

That is the key point. If comparing Miseq with Proton (the bigger brother of PGM) both of them are similar, then implicitly it recognizes that between "personal sequencers" Miseq is much better than PGM.

[aleferna]

I also don't understand the comparison between the Proton and the Miseq? Isn't Proton at the HiSeq 1000 price segment? Have the prices changed, I though after server, extra stuff the Proton/Chef/support ended up at 300k?? I when we bought our MiSeq the PGM was offered to us at $150k when it was supposly $50k.

Does anybody know an updated price table for 2013?

[steinmann]

I do not know about we are talking...

You confuse me do too

[victorsor]

Yes, but they should still be in the same ballpark which they are clearly not.


True, but recently released datasets show that this is not much of a problem any more.


60-80 million reads for the Proton is substantially more than the 15-17 for the MySeq. Error rates for the new chips are of course still a big unknown. Hard to say how smaller beads and a therefore weaker signal will impact data quality.

If the machine "star" of Life (not yet on the market) is only able to produce less than double output that Miseq (17 M. of reads x 500 bases= 8.5 GB this data at specifications 850 K/mm2 but actually 1000-1200 K/mm2 work well => up to 11 GB) and with a technology clearly worse: homopolymers, indels, error rate, Quality score (at least with PGM that is the machine that at today is available)... I do not know about we are talking...

[steinmann]

Guess my timing was not perfect but I did well enough

[MrGuy]

Maybe spun off to Roche? Or sold to Illumina?

[adaptivegenome]

Well, seems Thermo bought them.

http://finance.yahoo.com/news/thermo...120801466.html

Haha! Well good for them! I wonder what happens to Rothberg & Co.

[MrGuy]

Well, seems Thermo bought them.

http://finance.yahoo.com/news/thermo-fisher-buy-life-technologies-120801466.html

[steinmann]

Have sold my shares for now. Not sure whether the steep increase in price is fully justified. Reports over poor quality and shorter read lengths on the proton make me skeptical.

[aleferna]

Well, I would revise the traction of NGS if the price goes down to $100, old ideas are not necessarily bad. Turing had very bad decade old ideas that didn't get traction.

The stock market is really not were I measure a company's state. Its tough with the number of supercomputers driving artificially the price of the stock to a fibonacci number every millisecond, this has no relationship on real company progress, not anymore.

I was impressed by Illumina's decision to fend off the takeover, of buying Nextera, of responding to IT with the MiSeq.. I feel that the company is well managed. Life is too complex for me to wrap my head around.

Hi Aleferna, I am sorry but that vision is about 2 decades old and it's not gaining any traction. Just like ngs is not gaining ground in forensics and applied science.

Also, hope everyone sold their shares, life is officially way too expensive and that baby is slowly going to deflate.

[aleferna]

Let me first confess that I know I have no idea how to run a company, but engaging in these conversations even if I get a bashing like the one on top are really good learning experience. I want to thank BBoy for taking the time to respond to my many ramblings.

Some points that I want to defend:

I feel that in the last years (before ARM and the whole smartphone thing) Intel didn't get any real competition from AMD and that is precisely what left them vulnerable to ARM. Living around many people working at Intel factory, I heard rumors that they would hold back some new developments because they would make more money doing incremental steps than just throwing out the best they could make. Now, and again this is my naive perception, I felt that Intel kept AMD in business only to avoid anti-trust laws in the US. Also, I noticed that in the old times a CPU would be half as fast from one year to the other, the last time I exchanged my 3 year old computer for the top of the line i7 I didn't really see the difference even though I used the same version of Ubuntu. Yeap Intel has done A LOT, but I wonder what they would have accomplished with real competition, in that sense I called them "lazy", but of course I know there IS a reason why they never had competition, they we very very good.


"Yeah, that worked out for HP real well...": I thought it had? I honestly don't know, I just noticed that all printer makers suddenly dropped the printer costs and relied on expendables, not sure if it was a good strategy in the end but it certainly feels like nobody can compete with this strategy. I know printing is still like 20% of HP's market but that people are printing less nowadays because of email, there's also some macro-economic forces so it is really complicated to assess if the strategy worked, but I think it did keep them in the game as they would not have been competitive otherwise. Maybe if IT changes to consumables, Illumina would just do the same and they would have no gain. Still it would be really cool for us users though...

Is it me or is IT very good in creating Hype, 120k runs, but only like 300 submitted to SRA? If that is true for MiSeq, that means there are about 12M MiSeq runs no?

[BBoy]

Any updates on the contest, is Ion keeping up with the PII promise?

Not completely, they seem to be about six months off if you believe the latest statements from Rothberg at AGBT

http://nextgenseek.com/2013/02/ion-torrent-to-launch-ion-p2-in-mid-2013-jonathan-rothberg-at-agbt-2013/

Once you have bought a MiSeq or a HiSeq it becomes very difficult to through it away and by a Proton even if it has 10x more coverage.

That is only true in the RUO market, which is largely made up of NIH-funded purchases. In that market $100k is a lot of money. For a true commercial enterprise that is largely insignificant sum in the grand scheme of things. Ion Torrent likes to point to the semiconductor industry, and it is indeed useful to look there for what economies of scale look like. The big boys in that market depreciate sums in excess of $10B every few years. The major cost of switching platforms is not direct capital cost, but all the ancillary pain of re-establishing the surrounding infrastructure and the opportunity cost of establishing working relationships with a new supplier.

Also I have trouble understanding Ion, why are they trying to compete with Illumina's HiSeq, that monster is soon to be obsolete "mainframe". You should be building $10k self contained push-go sequencers for hospitals/farmacies/small labs???

Yeah, mainframes are nowhere to be seen anymore. Your phone runs on its own processing power. If there was ever a realization of Larry Ellison's "dumb terminal" vision, it is the "smart"phone. :-)

We need Ion to keep up the pressure in Illumina, otherwise she will turn into another lazy monopoly like Intel, unlikely to make any real advances because they have no real competition.

Your view of Intel is rather myopic. EVERY significant innovation in semiconductor process technology in the past decade has come from Intel, and everyone is copying them. The design side of the business is a somewhat different story with a complicated history, but "lazy" is not a word anyone who knows anything about Intel would use in describing the company.

My advice to Ion: you will never match Illumina at this pace, sell the PGM at cost, and make money on the consumables (printers/ink)....

Yeah, that worked out for HP real well...

Note: Yes I bought a single MiSeq when I really wanted to buy an Ion (I'm a disgrunted wanna be Ion customer). We would have bought the Ion if it was cheaper than the MiSeq, but it had no Nextera, we had to buy the Covaris and the OneTouch, much higher hands-on prep time and a huge footprint. Today if they offer a Proton at the same price as a MiSeq we would have to say no because we already have a MiSeq, we already made the investment... This is our own particular experience, but I know we probably not a rare case. I hope that Ion's manager has a plan, down here in the ground their future seems quite dim.

See, the ecosystem is usually far more important than any one particular component of it. The history of dominant enterprises shows that you cannot beat them at their own game, you have to shift the game. Today, at a very gross level the output of a P2 is very similar to that of a HiSeq: 200-ish bp scattershot reads, with perhaps lower consumable costs, but with a higher labor cost, perhaps a homopolymer issue that may or may not be important to many, and downside in the ecosystem (sample prep, informatics). If all IT can do is sacle output by making their chips bigger they are playing Illumina's game. (the cost of a YIELDING chip increases exponentially with area). If, on the other hand, they can make a 1B well chip and not have to rely on a HUMONGOUS 30mmx30mm die to do it then they are shifting the game and have a chance of eating the cake.

Speaking of that, does anyone have a sense how far IT can push well pitch? The latest PR statements from Rothberg imply around 1um pitch for the P2.

[Elcannibal]

Hi Aleferna, I am sorry but that vision is about 2 decades old and it's not gaining any traction. Just like ngs is not gaining ground in forensics and applied science.

Also, hope everyone sold their shares, life is officially way too expensive and that baby is slowly going to deflate.

[steinmann]

And in turn Intel is getting clobbered by ARM, not because ARM is better on the same metrics but the metrics people care most about shifted.

Funny that you say that. Intel happens to be my second largest long position

[krobison]

I am talking about a "kill" in the sense that Intel "killed" AMD.

And in turn Intel is getting clobbered by ARM, not because ARM is better on the same metrics but the metrics people care most about shifted.

Don't think the homopolymer issue is a showstopper. To me it looks like they have improved to a level that can be considered good enough.

Homopolymer issue might not be showstopper, but it is a royal pain. I'd be very curious if anyone trusts Ion data for finding frameshifts where it really matters (such as cancer samples), particularly in the face of low sample homogeneity.

Does anyone know if Ion has invested any in trying to fix this at level other than individual read? I.e., using suffix trees, kmers or other that look at whole set of reads? Definitely not something to attempt without informing user, but perhaps useful.