Michael Graham Richard

There’s a recent Economist piece that starts with: “What physics was to the 20th century, biology will be to the 21st”. Looking at what’s brewing these days, that’s not so hard to believe, except that biology probably won’t get a whole century; there’s a good chance that in a few decades nanoscale manufacturing and artificial general intelligence will take center stage. But anyway, here are two interesting developments that showed up on my radar lately:

The National Human Genome Research Institute (NHGRI) launched a public research consortium named ENCODE, the Encyclopedia Of DNA Elements, in September 2003, to carry out a project to identify all functional elements in the human genome sequence. The project is being conducted in three phases: a pilot project phase, a technology development phase and a planned production phase.

The pilot phase, which had for goal the analysis of a targeted 1% of the human genome (about 30mb of data, or 500,000 base pairs), is now over and the results have been published in the June 14, 2007, issue of Nature (you can see the press release here).

The next step is the “technology phase” which aims to investigate and develop new high throughput techniques and protocols suitable for the ENCODE project. Then there’s the final production phase which will “rigorously analyse the entire genome using the best methods and technologies identified in the first two phases”.

I haven’t yet seen estimates of how long the ENCODE project is expected to take - maybe nobody knows at this point - but it certainly will be a huge challenge, bigger than the human genome project on which it builds.

Links: The ENCODE Project: ENCyclopedia Of DNA Elements, UCSC ENCODE Project portal

pharmaceutical companies are reconsidering their pursuit of blockbuster drugs, as new technology permits the creation of niche remedies that target rare ailments or sub-populations of people suffering from common diseases.

That explains the $3 billion hostile takeover bid announced this week by Roche, a Swiss pharmaceutical firm, for Ventana Medical Systems, an American diagnostics firm. This year, Roche has gobbled up several diagnostics and genetic-testing firms making technologies that enhance the value of its targeted cancer therapies. The firm recently completed a $155m takeover of 454 Life Sciences [the company that sequenced James Watson's genome], which makes gene-sequencing technology and last week spent some $273m on NimbleGen, which makes technologies used in identifying the genetic causes of disease.

Roche is being drawn away from conventional one-size-fits-all drugs partly by the allure of the lucrative new markets being created by the development of “personalised medicine”. [...]

Anthony Farino of the consultancy arm of PricewaterhouseCoopers argues that such technologies as high-throughput sequencing, genomics and personal phenotyping, which were not available five years ago, are now transforming how drugs are discovered and tested.

Link: Beyond the blockbuster

Image: This is a Blue Gene /P node card. See full size image here.

The IBM Blue Gene /P and the SUN Constellation will soon go where no supercomputers have gone before: One petaflops and above. There is even a 3-petaflops version of the IBM Blue Gene /P planned.

“FLOPS” means “floating point operations per second”.

One megaflops is 1,000,000 floating point operations per second, one gigaflops is 1,000,000,000 and one teraflops is 1,000,000,000,000.

So one petaflops is 1,000,000,000,000,000 (aka one quadrillion) floating point operations per second.

To put this in context, the whole BOINC distributed computing network, with over half a million active computers, has a throughput of 562 teraflops (as of June 2007).

Like its predecessor, the Blue Gene /L, the /P is built around a tightly-integrated system-on-a-chip design that stresses low power consumption, a high number of nodes, and massive parallelism. Where the Blue Gene /L integrated these features around a dual-core PowerPC 440 processor running at 700MHz, the /P relies on a more advanced quad-core design that links four PowerPC 450 processors at 850MHz into a single package.

The one-petaflops version of the Blue Gene /P requires 294,912 processors and takes up 72 racks, while the three petaflops flavor requires 884,736 processors linked across a 216-rack cluster. Like the Blue Gene /L, the /P is capable of electrically isolating a failed node to allow the system to continue operation.

Images: Blue Gene /L, 65,536 dual-core processors, 360 teraflop peak speed, 1.5 megawatt power draw. Credit: Lawrence Livermore National Laboratory

the [Blue Gene /P] system is scheduled to deploy for the first time later this year at the Department of Energy’s Argonne National Labratory. Other deployments to follow include the Max Planck Society and the Forschungszentrum Julich research center (both in Germany), Stony Brook University, Brookhaven National Labs (American), and the Science and Technology Facilities Council in Cheshire, England.

And it’s not over: Blue Gene /Q, the next version of the Blue Gene architecture, is expected to reach 3-10 petaflops.

To see a list of the Top 500 supercomputers (those that aren’t kept secret, anyway), check out Top500.org.

Sources:

• IBM supercomputer Blue Gene /P breaks petaflop record
• IBM and Sun Microsystems hit ‘petaflop’ speed
• Sun debuts two-petaflop “Constellation” supercomputer
• BOINC Stats
• Blue Gene at Wikipedia
• FLOPS at Wikipedia

This one will be quick for a change.

• Simple switch turns cells embryonic

Research reported this week by three different groups shows that normal skin cells can be reprogrammed to an embryonic state [a.k.a. stem cells] in mice. The race is now on to apply the surprisingly straightforward procedure to human cells.

• Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls
• Serious diseases genes revealed

There is increasing evidence that genome-wide association (GWA) studies represent a powerful approach to the identification of genes involved in common human diseases. We describe a joint GWA study (using the Affymetrix GeneChip 500K Mapping Array Set) undertaken in the British population, which has examined approx 2,000 individuals for each of 7 major diseases and a shared set of approx 3,000 controls. Case-control comparisons identified 24 independent association signals [...]: 1 in bipolar disorder, 1 in coronary artery disease, 9 in Crohn’s disease, 3 in rheumatoid arthritis, 7 in type 1 diabetes and 3 in type 2 diabetes. On the basis of prior findings and replication studies thus-far completed, almost all of these signals reflect genuine susceptibility effects. [...]

Our findings offer new avenues for exploring the pathophysiology of these important disorders. We anticipate that our data, results and software, which will be widely available to other investigators, will provide a powerful resource for human genetics research.

Image: Picture of a young James D. Watson (born 1928). Public domain picture.

Not so long ago I wrote about about some of the controversy surrounding “celebrity” genomes: The first few individual genomes to be sequenced are mostly those of famous scientists. At first I didn’t think that was such a big deal, but I later changed my mind (see the post for details).

Well, now we learn that James D. Watson, a molecular biologist best known for his role in discovering the structure of DNA (the famous double helix) has received his genome (the $2 million DVD mentioned in the title of this post). It was sequenced in the last 2 months by 454, a company that is working on ways to read genomes more efficiently (and thus less expensively).

The $2 million and two months that it took to sequence Watson’s genome is a far cry from the more than ten years and $3 billion required for the Human Genome Project’s reference genome, released in 2003. Scientists ultimately hope to bring the cost down to less than $10,000, a target price that many believe will be the turning point in genomic medicine. At that price, many people could afford to have their genomes sequenced, and doctors could then use that data to give their patients more-personalized medical advice.

To get there and be able to interpret most of the genetic data we’ll first have to transcribe many more genomes and create databases that also contain people’s medical histories and personal characteristics. It should take a few years, but less than most people would expect. Exciting times!

Source: MIT Technology Review