There are only morning sessions today. It always seems that no one likes to talk the last day. Lots of people have left early to save a little money. But the ones who stayed got to listen to some pretty interesting talks.
The first set dealt with approaches linking human genotypes to phenotypes. Peter Park from Harvard discussed work analyzing the gene expression patterns of 23 patients with ALL and 11 with AML. They examined 7129 probes for 6817 human genes, using a scoring function that would help them find genes expressed in one group but not in the other. They found one gene, Zyxin, which was perfect. It was not expressed at all in any of the ALL patients but was found at high levels in all AML patients.
His nonparametric scoring function is pretty straightforward and I hope I can do it justice. All the patients in one group get assigned a 0. Those in the other get assigned a 1. They then determine the expression levels for each gene in each group. They order all the expression levels from low to high, with each of these still coupled with either a 1 or a 0. To get the score, the expression levels are shuffled around until all the 1s are back together, counting how many steps it takes. So any gene with a score of 0 means all the patients in one group express the gene at lower levels than all the patients in the other group. A score of 0 will be selective for one group of patients. Pretty neat. Zyxin had a score of 0.
Atul Butte from Children's Hospital in Boston had a very intriguing presentation. They examined the gene expression patterns from 6 patients that showed glucose intolerance but did not have diabetes. They used two Affymetrix microarrays examining 35714 ESTs. They compared the gene expression levels of the same patients on the two chips. Not too surprisingly, genes that expressed at low levels showed greater intra-array differences than genes expressed at higher levels. At low expression levels the inherent noise in the assay is a greater proportion of the values than at higher expression levels.
But what was really interesting occurred when they compared the fold-change in expression levels between pairs of patients on the two chips. Graphing the fold differences from each chip comparing, say, patient 1 and patient 2 showed some real problems. In an ideal world, you would expect that a gene that expressed, say, 5-fold better in patient 1 on the first chip, would also express 5-fold better on chip 2. Graphing all 35714 ESTs should give a straight line. In a perfect world. In the real world it gave a circle. That is, some genes were 10-fold higher in patient 1 on the first chip and 10-fold lower on chip 2. Others were the reverse.
The pairwise comparison of every 2 patients revealed the same thing. So Butte worked out several approaches to statistically identify which ESTs had a really significant difference in expression levels. Here is a link to the paper. It was pretty interesting and a nice presentation of a difficult problem.
The
last session looked at Phylogenetics in the Post-Genome Era. This one had
several discussions on developing better algorithms for generating phylogenetic
trees. While useful, all of these were pretty specialized 
(i.e.
looking at Drosophila species or HIV-1 variants). Eben made it though the
last talk. But Randy and I were dozing off well before that one ended.
The talks were over at 12, so I took the afternoon off and went out on a sailing canoe to look for whales. The humpback whales migrate into Hawaiian waters this time of year. They give birth to calves and mate. Maui is the hot spot for tourists to go whale watching. I know, having done it several times. There are a ton of boats that just follow the whales around in the Maui Channel. It is pretty crowded and, even though most tours try to obey the non-nuisance laws, I always felt the whales deserved a little privacy.
Well,
the Big Island has just as many whales, probably trying to get away from
all the boats around Maui. So, I went out on this authentic Hawaiian sailing
canoe. Made by hand from hawaiian koa trees by the captain, Casey Cho.
With fishing lines trolling behind us, we ventured out, and quickly came
upon a mother and her youngster slapping their pectoral fins on the water.
We watched for half an hour and then came upon another 2 whales who were
playing with each other by breaching. We sat out there, on this wooden
canoe, a couple of miles off shore, in the sunshine, just rolling with
the waves, eating pineapple and watching whales. It was a beautiful coda
to a wonderful meeting.
I'll be flying home tomorrow and getting in late. This turned out to be a very exciting meeting and I hope these reports have been useful. Let me know how you liked them. I really had a great time writing them. I learned a lot about digital photos and movies. And about remote access. And a little bit about alignment algorithms, torsional stress of DNA, the draft sequence of the human genome, high performance computing, the generation of genetic networks, and the problems associated with analyzing array data. But I am looking forward to getting home. Sleeping on hotel beds is never as nice as one's own. Aloha.