A Bloomberg Terminal For DNA Sequencing

Last Thursday, Knome, a small, Cambridge, Mass., startup, announced that in the fourth quarter it will begin selling a $125,000, 600-pound supercomputer to hospitals to allow them to analyze the data contained in one human DNA sequence in as little as a day, at a time when the cost of that data is dropping toward $1,000.

The idea reminds me of nothing so much as the Bloomberg terminals that you find in financial institutions. It's far more expensive, of course, but this new device is essentially a realization that sometimes experts want particular types of data a few steps away, putting the DNA analysis service business Knome is already running in a box.

Knome's shift presents a snapshot of DNA sequencing at a moment when it is finally making its way into the clinic, even, in some cases, being paid for by insurance companies, and a raft of start-ups are trying to push their way into what seems to be a new medical ecosystem at the same time as the last round of entrants are struggling (see: Pacific Biosciences of California) or getting bought out cheap (Complete Genomics, acquired by China's BGI for just $117 million.)

Knome started out in 2007 selling vanity genomes, founded by Harvard professor George Church after he was approached by people willing to pay an arm and a leg to get sequenced. The initial cost: $350,000. The first customer was Dan Stoicescu, a Romanian-born pharmaceutical entrepreneur. In 2010, Knome sequenced Ozzy Osbourne and his wife, Sharon; in 2011 its scientists were featured in a PBS documentary comparing the genomes of Henry Louis Gates, director of the W. E. B. Du Bois Institute for African and African American Research at Harvard, and his father.

But as the cost of DNA sequencing dropped, Knome found its business focused more on researchers who wanted help analyzing DNA samples. At the end of last year, this business accounted for about $3 million in revenue bookings, according to the company.

In January, Martin Tolar, a former Pfizer executive, joined Knome as its new CEO. Tolar's plan was to focus more on drug companies, who he says will account for some $4 million or so in bookings this year. Knome has also expanded from its initial offering, which just helped academics with analysis, to a more full-service option, with will bring total bookings to $9 million or so this year. (Knome doesn't recognize all the booked revenue at once.) As revenue has increased, Knome has been analyzing a lot more genomes. It has booked 3,000 genomes, and finished analysis of 1,500 genomes.

But many customers, particularly hospitals, wanted to be able to do some analysis within their own walls. "People need accurate interpretations that they have control over," says Church, the Knome co-founder. "They don't want to be forced to send stuff out. That's an awful lot of data that's quite sensitive."

That's where the idea of selling a computer – essentially a decked out server with proprietary software – came in. The machines are monsters, with 2.4GHz 8-core/16 thread Intel® Xeon® E5-2665 processors, and between 18 to 54 terrabytes of useable disk storage. And one will be required to man each DNA sequencer, and they'll be used on top of the computing solutions already sold by the makers of DNA sequencing gear.

Right now, there are about 2,000 next-generation DNA sequencing machines out there. That vast majority are made by Illumina (total sales: $1.0 billion) and have a list price of about $700,000. A new entrant, called the Ion Proton, will retail for $120,000 and is made by Life Technologies (the underdog in sequencing, Life has sales of $3.8 billion in other businesses.) A likely estimate, Knome says, is that there will be 1,600 Illumina machines and 600 protons by sometime next year.

One KnoSys machine will be needed for each DNA sequencers whose data scientists want to analyze with it. Some institutions, such as Knome's neighbors at the Harvard/MIT Broad Institute, have their own informatics solutions that are already pretty great. Others may not need that firepower at all. But Tolar says he thinks its possible to sell machines to 10% or 20% of the installed sequencer base. That would be a market of $25 million or more – twice as big as Knome's current bookings.

Knome is hardly the only company trying to become the "big data" entrant for DNA sequencing. There are other entrants, like DNANexus, which offers a cloud-based solution, and Foundation Medicine, which focuses specifically on cancer. But its evolution offers a glimpse into how fast things are changing. Just a few years ago, you could gather all the people whose genomes had been sequenced in one room. Now there have been tens of thousands such individuals. The big question now is how many there will be in a few years – and how much of a market opportunity there is in figuring out what all that DNA code means.