ISMB 2007 Vienna -- Part II Interesting Talks

I'd just like to comment on some talks that I also found particularly noteworthy or interesting.

Atul Butte gave an interesting talk on nosology. No, that's not the study of noses, but rather the classification of diseases. What Butte has done is cluster gene expression patterns from various diseases. And the results were surprising. For example, he discovered that cervical cancer expression clustered with that of an autoimmune disease andmuscular dystrophy clustered with some forms of heart disease. These connections were unexpected and may lead to better understanding of these diseases. I liked the idea (even though I am not normally a disease person) because I found the idea analogous to the way unexpected phylogenetic relationships can be informative.

Christian von Mering talked about phylogenetic analysis of samples from metagenomics projects. He is interested in this problem because contrary to what was generally believed prior to environmental sequencing, specific microbial species (& higher taxonomic levels) are not found everywhere it would be possible for them to live -- in other words, microbes have meaningful biogeography, just like plants and animals. Von Mering and colleagues have created an interesting pipeline for the interpretation of metagenomic data. Rather than try to analyze each of the millions of reads in detail, his method first identifies standard phylogenetically informative marker genes, then adds them to existing high quality alignments, and uses a custom phylogenetic program to test all possible phylogenetic positions on a reference tree. Both the limitation to markers and the custom phylogenetic component make this pipeline much more efficient than the typical phylogeny based methods.

Haipeng Li talked about a new maximum likelihood method for inferring positive selection and demographic history from chromosome-wide SNP data. His test case was Drosophila melanogaster and according to his model, the European population split off from the African lineage 16,000 years ago and underwent some sort of bottleneck. Current low levels of X-linked diversity (as opposed to the autosome) in the European population suggests a large excess of males in this population. This latter assertion was questioned by several members of the audience as it doesn't seem to be congruent with empirically measured numbers of males in the wild. Still, I find such studies fascinating. I would love to do a similar demographic study using bacteria, but I doubt we have enough data, even for things like E. coli or Bacillus anthracis.

And Alissa Resch, who works at the NCBI (just down the road from JCVI), also gave an interesting talk about positive selection. Odd that I'd have to travel thousands of miles to hear it though. Resch developed a new statistical test for positive selection that uses the rates of synonymous substitution in nearby intronic regions as a background. This allows detection of positive selection even at synonymous sites. She then applied it to orthologous gene pairs in mouse and rat.