Friday, February 08, 2008

More notes from Marco Island/ AGBT

Some notes on talks here:

My favorite talk yesterday morning was David Cox from Perlegen. He had as usual some good one liners including "Everybody and their mother is doing this so doing this is not so novel. What is novel about it is that it worked." I should add that David Cox helped shape my career indirectly in many many ways. When I was a PhD student at Stanford, I got into genomics in part by teaching a course with David Botstein, Rick Myers and David Cox. When Craig Venter offered me a job at TIGR in 1998, I was not sure if moving to a non university was a good idea or not. So I asked many people for their opinions. Some said "You must do an academic post doc or you will never get a faculty job" I pretty much knew to ignore those folks. Cox gave the best advice. He said as long as I published things while at TIGR, it would not hurt me in any way. It probably would help. And so I took the job. And no doubt that was a great career move.

Other talks that were good were one by Joe Ecker, who discussed methylation in Arabidopsis and one by Andy Clark.

I skipped out on some of the lunch time to finish my talk for the PM session and also worked on my talk in the back of the room during the other PM talks. The PM session was on metagenomics and the most pleasing thing was that David Relman did not show up and he was replaced by Peter Turnbaugh from Jeffrey Gordon's lab. Now - I wam not saying it was good that Relman was not there --- he usually gives smashingly good talks. But Turnbaugh, a PhD student, stepped in as pinch hitter and gave a great talk on gut microbiome studies, really setting the stage for the whole session. I do not know if he was nervous stepping into a session like this but it did not show if he was. He certainly seemed relaxed when he said "Thanks to Dr. Relman for getting stuck in Chicago"

Forest Rowher gave a good talk on metagenomics and pointing out that viruses still get ignored in this field relative to their likely importance in communities. I have written about Forest before so I am going to discuss the other talks more ... but if you have not heard him talk before try to find a way. He has a VERY different perspective on genomics and metagenomics than most of the people doing it. And he is dead right about the need to do more work on viruses.

Garth Ehrlich gave a talk on "bacterial plurality" and why he thinks gene content variation within communities of microbes in biofilms is important. His data certainly seemed solid and he showed some results that call into question the claims that some aspects of the "pangenome" hypothesis (he showed that the total number of genes in the Steptococcus strain collection does seem to level off after sequencing ~ 30 genomes and thus that the number of genes is not infinite as some people have suggested). So I liked some aspects of his talk. But he did make some evolution statements I found disagreeable (for those who care about the nitty gritty - he showed a cluster diagram of strain similarity and then used the position of strains within the cluster diagram to reflect relative branching order and historical patterns. A cluster diagram is a bad thing to use and one should use a phylogenetic tree for this. In addition he implied that one could make a genome-phylogeny from gene presence/absence information that would be more robust than a standard alignment phylogeny. This is not a reasonable thing
in my opinion --- gene presence/absence patterns tend to end up grouping together unrelated lineages that have separately undergone gene loss. I just do not understand why people so badly want to not use alignments to build trees). Anyway - overall many of the things he said were interesting but I find certain non-evolution evolutionary analyses really grating.

Anyway - I was going to ask him a question after his talk about this, but then decided that, since I was talking next, getting into an argument with him just before my talk might seem lame. So I passed on the question. And then I gave my talk on the need to fill in the tree of life in terms of genome sequencing projects. I discussed a project we are just wrapping up that was part of the NSF "Tree of Life" program in which we sequenced genomes of eight bacteria that are from phyla that at the time had no genomes available. And then I talked about a new project I am coordinating at the Joint Genome Institute in which we are sequencing 100 genomes to really fill in some of the bacterial and archaeal tree. Next week I will post more about this project but I note - this is not done to study the tree of life per se. It is being done because if we have reference genomes from across the tree, all of our genome analyses of other systems and of metagenomes get better.

After dinner and some shell cllecting on the beach, there were evening talks and I went to the informatics session. Some of the talks there were good but the best thign I saw there was someone (I think Ben Blackburne) saying his slides were going to be on something called slideshare.net. I had never heard of this and checked it out and it seems pretty cool. I may use it in the future ... but gotta go off to other things.

2 comments:

  1. slideshare is great. Jean-Claude, Duncan Hull, myself and others use ti quite regularly to put up publicly available slide decks. You can even add voice after the fact if you want to.

    ReplyDelete
  2. Thanks for sharing about the slideshare.net, it looks really useful.

    ReplyDelete