The Motley Fool Previous Page

The Promise and Peril of the Mass-Market Genome

Alex Planes
January 11, 2012

The relentless march of genome sequencing technology continued this week, right on target. When I wrote about the race to the $1,000 genome late last year, I had no idea that the two largest sequencing companies would so soon unveil machines capable of offering just that, but I knew it was coming. Now that it's here, what does it mean for the future of medicine -- and the future of the planet? More than you might think.

The day the walls came down
Jan. 10, 2012, is an interesting day in medical history -- Life Technologies (Nasdaq: LIFE  ) announced that $1,000 sequencing is finally coming. Life Tech's planned release date for a full-genome sequencing version of its new machine is around the end of this year. Less noticed was rival Illumina's (Nasdaq: ILMN  ) newest sequencer, which didn't grab similar headlines because it failed to name its sequencing price. However, Illumina's model is projected to put the $1,000 mark "within reach," and could make up some ground with raw processing power.

Since Life Tech's announcement so closely conforms to my earlier projections, let's revisit them, and this time I'll stretch the cost curve out all the way to the end of the decade:

Sources: National Human Genome Research Institute and author's calculations.

Race to the bottom
Life Tech's price point, should it hold up, will maintain the rapid decline in sequencing costs that began at the end of 2007. If that trajectory continues, the $100 genome won't come in 2016, as I originally predicted by conservatively tracking the orange line. It'll get here by mid-2014. By 2016, the faster pace Life Tech's now staked itself to will bring full-genome sequencing down to a cost of just $3. By then, a real medical revolution should be under way.

The positive picture
The thought of buying your decoded genome for the price of a half-decent hamburger seems absurd in its optimism, but there's yet to be a significant slowdown in this breakneck progress. At this point, it seems far more important to prepare for the inevitability of the "everyday genome" than to continue dismissing it as the expensive curiosity it will soon cease to be. And that means accounting for the promise -- and also the peril -- of mass-market sequencing.

One clear benefit to broader understanding is better drug targeting. Amgen and Pfizer's (NYSE: PFE  ) Enbrel costs $26,000 a year, yet only benefits half the arthritis sufferers taking it. Knowing a patient's unique genomic variations could help identify why they don't respond to certain drugs, leading to smarter prescriptions. It's also likely to lead to more efficient drug research and development, but this is more contingent on the availability of large genomic databases. That would help companies like Pfizer that face a looming patent cliff.

Incredible implications
Understanding how individuals differ on a genetic level would very likely aid doctors in diagnostics, particularly in the discovery of genetic diseases. Today, many such diseases are often diagnosed individually through a hodgepodge of different tests. Sequenom's (Nasdaq: SQNM  ) Down syndrome test, which is processed on Illumina's machines, is one example. Looking forward, an ambitious National Institutes of Health prediction for the year 2020 anticipates that all fetuses will be screened by seven weeks for more than 200 genetic disorders, which can then be repaired by the necessary gene therapy.

The NIH's 2020 predictions reference genetic analysis at nearly every possible opportunity. On the back of better genetic knowledge, they claim, we'll be able to:

  • Repair abnormalities of the eye.
  • Develop individualized drug treatments for diseases (both chronic and acute).
  • Better understand aging and diseases caused by it.
  • Control or cure alcoholism.
  • Identify and treat hereditary deafness.
  • Control or treat drug addiction, including treating the effects of drug addiction.

All this, incredibly, was predicted in 1999. The directors responsible for these predictions would have known only a very immature technology -- the first full, individual human genome wasn't sequenced by J. Craig Venter's Human Genome project until 2007. Now, the numbers say that a full genome sequenced in 2020 will cost somewhere between nothing and the price of a convenience store hot dog. Will sequencing become that cheap? I don't think so -- but it will be cheap enough to force us to answer some difficult questions.

The measure of a man
What can we do with the genome today? Beyond a few screens, not much. Genetic information gives rise to an astounding variety of individual differences. The differences between two siblings -- or the spread of biometric scanners based on unique, minute differences in the human eye and fingerprint -- offer straightforward lessons in the seemingly limitless capacity for genetic variance.

We can sequence one person's genome and identify risk factors that may make him more likely to get cancer or heart disease, but still can't deal with these problems much past healthy-living preventive measures. Say your prayers and eat your vitamins, so to speak. If you got a full genome screening today, it would probably reveal 200,000 individualized variations that have never been seen before. A Life Tech sequencing of one person turned up almost 4 million single-base (half of a base pair) variations. That's a lot of unknowns.

Exponentially exponential
There may be ways to overcome the variance challenge, but it won't be easy. Comprehending and interpreting so much variety adds layers of costs that aren't likely to drop quite as quickly. One of the biggest attractions of Life Tech's new machine, besides its low cost, is a purported ability to shrink processing time to a day from the week-plus times required by current Illumina machines. But after processing comes analysis. Current algorithms add days of extra time to the process. After analysis, the results need to be interpreted by genetic counselors, who may need hours to explain to patients what those results mean.

Let's say we cut total processing time to a few minutes. It's certainly possible. Algorithmic performance is in many cases far outpacing the advance of its underlying technology. One model that could have been solved in 82 years beginning in 1988 was later solvable, in 2003, in one minute. This 43-million-fold improvement was attributed to a thousand-fold increase in processing power and a 43,000-fold improvement in the algorithms used. An algorithmi