Re: Genomics [was Re: MEDIA: Celera profitability]

Robert J. Bradbury (bradbury@www.aeiveos.com)
Mon, 27 Sep 1999 10:02:11 -0700 (PDT)

On Mon, 27 Sep 1999 k_aegis@mindspring.com wrote:

> Today's Washington Post Business section has an article regarding the
> potential return for investors as Celera prepares to complete its
> human genome research.
>
> http://www.washingtonpost.com and then scroll down to 'A Gene Dream'
>

The actual URL is:
http://www.washingtonpost.com/wp-srv/WPlate/1999-09/27/031l-092799-idx.html

An interesting article. Mostly accurate. Of couse the devil is in the details. Celera is in a race with several government funded labs to publish genome sequence. To Celera's credit is the fact that they have finished sequencing the Drosophila genome (at 10-11x redundancy) and this sequence will be published in late October. However, with regard to the human genome, the government funded labs and Celera both are on track to publish significant amounts of the human genome next year.

What was really interesting about the Celera talks at the TIGR genome conference was that fact that once they have a prototype mamamalian genome (human) they can drop the sequencing requirements (for random short fragments) from 10-11x down to ~3x. So following the human genome, the mouse, etc. genomes should follow rapidly.

The fundamental problem with large genomes is the difficulty of ensuring that you have things reconstructed in the right order given inversions, duplications, repetitive elements, etc. One of the presentations at the TIGR conf. lead me to believe that the Drosophila genome is more difficult than the human genome (more repetative sequences). So though the Human genome may be bigger (requiring several months of computing time for a 1+ TeraOp machine), its complexity is lower, so the assembly process is likely to be fairly straight forward.

Clearly the government scientists and strategies have been scooped in this affair. However the credit should go more to Marc Adams who clearly knows how to assemble a high throughput production facility and Eugene Myers who clearly understands the mathematics and programming requirements.

The *fundamental* problem that Celera faces is harvesting enough information from the project that they can come up with a few hundred unique patents to justify the investment. Can they do this? Probably. They have to strech out quite far beyond simply sequencing genes and take it into the realm of function where they can solve the "unique" and "nonobvious" criteria for patentability.

However, Human Genome Sciences appears to have made this transition, so I expect that Celera may as well.

Robert