Notes from KINCON 2005, biotech thoughts
Some notes from day one at KINCON 2005 at the Palo Alto Crowne Plaza. Today’s sessions were technology-focused. Although this has traditionally been a Korean IT-related conference, and mostly chips and displays at that, the biotech presentations were the most interesting.
The first session was on wireless technology, mostly aimed at services for mobile phones, such as ringtones and games. Korea is a good place to try launching these services, with 76% wireless penetration rate, and 90% of the handsets capable of running games and multimedia. My observation — it’s hard to do much in this space with the mobile operators trying to extract fees from the customers and 3rd party service providers (in order to pay back their spectrum license fees). At least two of the speakers commented that most of the fee paid by the customer is to cover the billing costs.
The second session was on consumer semiconductors. Everyone in this session is a fabless design house. The Xceive presentation on their fully digital TV tuner chipset is interesting. This is a completely silicon RF-to-baseband system, which apparently doesn’t require any external filters or shielding. They didn’t mention how much their first products were going to cost, but getting everything onto silicon means the cost is likely to trend downwards, rather than staying put as the existing analog tuner cards have. It also means that the physical packaging is much smaller, so it could go in a PDA or phone or be used to add a video input feature to an existing digital device. There are already a number of cheap video player widgets starting to turn up, sort of like video MP3 players, and having a silicon-only solution for RF TV-in is going to enable a lot of interesting combinations.
The third session, on biotech, was the most interesting. Fan Hsu from the UC Santa Cruz Genome project gave a general overview of proteinomics and functional genetics. Lewis Williams from Five Prime spoke about their process for screening thousands of candidate secreted proteins against specific cell functions, vs the old method of testing a single protein at a time. Stuart Kim from Stanford spoke about the need for a systemic view of protein function and the possibility of applying a broader engineering approach to modelling clusters of related gene expression. His project uses Affymetrix GeneChips or something very similar to test aging-related gene expression. I got the impression that it was something like the work being done at Perlegen a few years ago when they were doing their sequencing project, in that they were generating huge quantities of data, without a good method for organizing and modelling the results. Each test chip returns something like 5000 columns x 20,000 rows of results. The last talk was by Christopher Ko at Samsung Advanced Institute of Technology, who hopes to take GeneChip-like technology from a research environment down to something more user friendly, perhaps even to the point where an individual could run a test at home, sort of like a home glucose monitor.
I noticed that a lot of people left before the biotech session started, perhaps because the usual audience for this conference was interested in “InfoTech” and not so much “Bio”. My observation here — the genomic and proteinomic fields are just reaching the point recently where the information technology and systems engineering can become really useful. Another observation, though — pharma and health care product development are massively capital intensive and yet have very high risk per investment. The size of individual investments makes it difficult to place a lot of bets and plan on the portfolio paying off. On the other hand, it looks like there should be many opportunities to make a contribution toward advancing the state of the art, since the availability of data and computational tools is relatively new.
This evening I saw the announcement for Maxtor’s latest generation of desktop hard disk drives, which will be 500GB per 3.5″ unit; similar units from Seagate and Hitachi are expected in a similar time frame. The entire human genome database apparently requires something like 3GB, and the annotations from various research projects bring the total up to 8-9GB at the moment. So the current and future generations of desktop (and notebook) computers will have more than enough raw storage to handle the data sets. Doing something useful with the data is another problem altogether, but the size of the genetic/proteinomic database is relatively finite — it isn’t going to get exponentially bigger — and the computational resources continue to get exponentially larger/cheaper/faster. Something good has got to pop out of this somewhere…
See also: Korea’s plans for Ubicomp City, Korea becomes the largest foreign investor in India
Tags: kincon
