watson.jpg

“Life, with its rules, its obligations, and its freedoms, is like a sonnet: You're given the form, but you have to write the sonnet yourself.”

- Mrs. Whatsit

― Madeleine L'Engle, A Wrinkle in Time

IBM’s Watson is good at R&D, and could be good at Clinical R&D, but I would be surprised to see it used for Clinical Standard of Care. Why?

It would be like asking the super computer that runs the wind tunnels at Airbus or Boeing to run the aircraft in flight. Neither system is better or worse than the other, just different tools for different jobs. There is not a one size fits all genomics solution. We see this already in the current funding activities – genomics wholesale solutions are taking a far back burner to specific solutions to specific problems – see the current focus on Cancer and Personalized Therapeutics (See this nice article at Nature: http://www.nature.com/news/use-of-personalized-cancer-drugs-runs-ahead-of-the-science-1.18389)

There are really three main types of bioinformatics activities out there. Let’s start with the easy ones first:

  1. The most capable, most generic solution. This is a fantastic fit for Watson. This is the pure R&D play – what Watson was made for. There is no better technology out there right now than what Watson provides. These are large funded projects, not modeled on specific unit pricing.
  2. The most focused, most specific solution. This is not a great fit for Watson. This fits very closely with the traditional pharma diagnostic and therapeutic model: What is the cheapest, fastest, lowest risk test for disease X? This is not a great fit for Watson. Technologies like Bina (recently acquired) are targeted at this space – “appliances”, “turn key” and other phrases are what to look for here. Here is where the unit price model makes sense – pricing in the tens of dollars are the target here.

The last option is very specifically called “everything between 1 and 2” and it has a MASSIVE amount of variability – from applied agriculture research, to longitudinal studies, large scale analytics of chronically ill patients… There’s a lot here. In some cases Watson would be a good fit, but the vast majority of organizations are missing two things from making significant use of something like Watson:

  1. Trained staff – let’s be honest – this is HARD and we don’t have nearly enough trained people around the world (esp. in Asia) who have the skills to build and run these systems.
  2. Checkbook – let’s be even more honest – this is EXPENSIVE. Until Watson is in the cloud and at a pricepoint only slightly higher than Amazon’s AWS will this cease to be a limiting factor.

Given all of that, who are the “competitors” for Watson?

  • Sequencer Manufacturer platforms (Cloud or Appliance based)
  • Companies like Illumina, Ion Torrent (Life Technologies) are extremely well positioned to move up the value chain; Helix, funded by Illumina, is a perfect example of that move up the value chain.
  • But in order to do so they need to also improve their foundational reference architectures, investing beyond out of date, open source stacks which are fine for R&D but don’t scale for the “big time”
  • Growing Enterprise Cloud Platforms
  • AWS – Seriously! One of the best kept secrets in the global genomic industry is that the vast majority of genomics work (even clinical) is being done on Amazon Web Services; if Amazon ever decided to go deep into real time medical businesses, they’d grow a multibillion revenue stream and be one of the companies to “beat” to get to scale.
  • Google – if the market goes unchecked, in 10 years they could end up winning because they’ll write the check to buy the market; today Broad Institute of MIT and Harvard and Google Genomics are exploring the need for computing infrastructure to store and process enormous datasets, and tools to analyze the data. Google is not standing still.
  • 3-5 Tier 1 & 2 Academic groups including some in the Ivy League who have interesting hacked together platforms by the emerging talent coming out of universities and colleges with MDs – and MBAs
  • Labcorp/Qwest – If either decided to build something of high quality, they’d win since they already have the biggest footholds in the market. Labcorp is already beginning to label itself as, “a leader in genetic testing”.
  • A new company that will burst on the scene with the best team of visionaries, scientists, commercial officers, and strategic investors who understand the economics of lower cost to build better platforms, more integrated and fueled by increasingly less expensive compute and networking.

In my third post in this three part series, I’ll share some of my observations on the challenge of the business of Genomics.

Read the first post here: Where Watson Fits into the Emerging Global Genomics Ecosystem