Wednesday, May 11, 2011

Ion's Growing Pains

A recent In Sequence article indicated that Ion Torrent is enjoying strong initial sales.  This bodes well for continued evolution and improvement of the technology, as LIFE will continue to smell revenues and opportunity.  Ion has announced a number of improvements, but most aren't scheduled to arrive until the near future.
The challenge is for LIFE to keep executing on their plan.  Already some issues have arisen; my own Ion experiment at a service provider is stalled due to a back-ordered template prep reagent (two weeks and counting!). This is a key reason to do pilots on emerging technologies with non-critical (but interesting) samples; I was bitten last fall by another backorder bug (that time, the paired end SOLiD reagents).  Of course, this time it is even more complicated, as Ion is making a major change to both the underlying kits and to the software to process the data.  It will be worth it if I get results anything like Ion's provided E.coli 314 dataset, which has about 4.5X the data of the original 314 chip spec.

But will this same problem of rolling out improvements persist?  I'm really hoping to see the link to the 316 chip become active in the web store by next month.  The 316 sports about 5X the number of sensors as the 314 and will presumably give 5X the number of reads.  If it can really deliver the same quality of data and that E.coli dataset is not horribly cherry-picked, then the 314 could debut over 2X over its specification.  That would be truly impressive and definitely affect designing experiments for the platform (in a good way; more data is always better!).  
Similarly, the OneTouch template prep instrument should appear by the end of the summer and could be a big driver of sales for the whole system; if you value your staff's time at $1.5K+ a day (a reasonable ballpark figure for the fully-loaded cost of an employee in Boston or similar U.S. cities), a $5K instrument that saves about 1/4 to 1/2 a day per run is an easy sell.  But, if the first boxes don't show up on time or don't perform to spec, then the opportunity to steal a lead on MiSeq will erode.  Plus, a big selling point of MiSeq is that the technology is well-tested, and delivery hiccups by Ion will only reinforce that impression.  Ditto for the 318 chip; delivered on time it will give Ion a performance edge over MiSeq.  Conversely, late delivery will mean lost instrument sales and worse the lost opportunity to sell reagents to those customers.
On the informatics side, there were a number of useful comments on my analysis. In particular, Nils Homer pointed out some tuning I could do on TMAP parameters (which I haven't actually run yet).  Nick Loman has generated an interesting take by using the same dataset to try out different assemblers.  Nick shows how much you can do in the current environment, but also suggests a strong need for Ion-specific or Ion-tuned assemblers.  Also, the fact his lab hasn't run their Ion Torrent yet is telling, reinforcing the notion that Ion customers are being forced to pause by protocol shifts and reagent shortages.  Such software won't get written unless developers are motivated and have access to test datasets.  In many cases this will be because the developers are in labs with PGMs, but to really grease the skids Ion needs to put more datasets out in the open.
Perhaps least of Ion's worries is the lack of upstream preparation kits and protocols, as there is plenty of interest in the existing applications of fragment sequencing, amplicon sequencing and library QC.  Still, it is surprising that LIFE hasn't retrofitted some of the existing SOLiD kits for the Ion; in theory it should just require changing a few oligos.  Alternatively, for some applications a few "cheater" oligos could convert another platform's library into an Ion library, at the cost of losing 20-30 bases of read to the other platform's universal sequence.
Ion has a lot of upward potential and is already demonstrating an ability to beat performance targets.  But delivery targets are critical too, especially when eager customers are left champing at the bit.

7 comments:

Anonymous said...

Thanks for your post, Keith. We’re very excited by the enormous demand for the Ion PGM™ System, and at the rate we are placing instruments, the PGM will have the highest installed base of any next-gen sequencing instrument within the next 12 months.

We currently have an ample supply of template kits available to our customers and appreciate your input and efforts to further enhance our platform both technologically and commercially.

Sincerely,
The Ion Torrent Team

Anonymous said...

Thanks for your post, Keith. We’re very excited by the
enormous demand for the Ion PGM™ System, and at the
rate we are placing instruments, the PGM will have the
highest installed base of any next-gen sequencing
instrument within the next 12 months.
We currently have an ample supply of kits available
to our customers and appreciate your input and efforts to
further enhance our platform both technologically and
commercially.
Sincerely,
The Ion Torrent Team

Anonymous said...

Typical.

Anonymous said...

Of course the Ecoli run is cherry picked, and with Qscores that low, those are some sour cherries.

I'm wondering what the false positive rate for SNP calling is going to be when you factor in those low scores. This is particularly a problem when you have multiplexed samples and the coverage per sample is reduced... My guess is it will equal more runs, more costs and expanded project times.

Also, if they have an ample supply of kits, why can you get one?

Muttley said...

I have been reading your posts for a while, and one thing I cant get my head around is the astonishing lack of objective skepticism being applied to the claims ion are making around their platform. The only data that you have seen, despite over 5 months of the system being on the market, and apparently tens or hundreds of system sales, is an e coli set provided by ion. Don't you find this odd? You are a well respect computational biologist, well linked in the community, and yet you can't lay your hands on an independent data set. Despite this, you are willingly parroting the marketing claims being made for chips that don't yet exist commercially. You are the best form of free marketing they could ask for. And yet, on the seqanswers site, you are found griping about the lack of access to technical information on their protocols. You tout the strengths of ion for amplicon sequencing, but don't point out that their short read lengths and current sequencing approach mean that at best you will read 40-50 reasonable quality bases of real sequence. They need to be at least at 200base reads to become relevant here.

Like it or not, people are going to be influenced by your opinion, and you have some responsibility to apply the same standards of objectivity to the new products being released in this market as you would to your own experiments. The default should be "don't believe the claims until you have seen independent data to support them".Otherwise, you are simply facilitating the spread of unsubstantiated self-promotion.

If you apply the same level of skepticism to other platforms on the market, then PACbio will be sequencing human genomes for $5k within the year, 454 have had 1k reads for the last 3 years, and the 150th base read on illumina is as good quality as the first.

"supposing is good, but finding out is better" mt

Muttley said...

I have been reading your posts for a while, and one thing I cant get my head around is the astonishing lack of objective skepticism being applied to the claims ion are making around their platform. The only data that you have seen, despite over 5 months of the system being on the market, and apparently tens or hundreds of system sales, is an e coli set provided by ion. Don't you find this odd? You are a well respect computational biologist, well linked in the community, and yet you can't lay your hands on an independent data set. Despite this, you are willingly parroting the marketing claims being made for chips that don't yet exist commercially. You are the best form of free marketing they could ask for. And yet, on the seqanswers site, you are found griping about the lack of access to technical information on their protocols. You tout the strengths of ion for amplicon sequencing, but don't point out that their short read lengths and current sequencing approach mean that at best you will read 40-50 reasonable quality bases of real sequence. They need to be at least at 200base reads to become relevant here.

Like it or not, people are going to be influenced by your opinion, and you have some responsibility to apply the same standards of objectivity to the new products being released in this market as you would to your own experiments. The default should be "don't believe the claims until you have seen independent data to support them".Otherwise, you are simply facilitating the spread of unsubstantiated self-promotion.

If you apply the same level of skepticism to other platforms on the market, then PACbio will be sequencing human genomes for $5k within the year, 454 have had 1k reads for the last 3 years, and the 150th base read on illumina is as good quality as the first.

"supposing is good, but finding out is better" mt

Anonymous said...

There are some independent data sets out there, but from people who got the Ion Torrent systems early. Unfortunately, Ion Torrent is using all the legal muscle they can muster to suppress publication of such data.

The data is not terrible, just not as good as Ion Torrent wants to paint things.