Keynote Speaker - Dr. David Haussler, Howard Hughes Medical Centre

Identifying functional elements in the human genome by tracing the evolutionary history of the bases: a key challenge for comparative genomics

Additional contributors: Mathieu Blanchette, Adam Siepel, Krishna Roskin, Ryan Weber, Mark Diekhans, Francesca Chiaromonte, Ross Hardison, Jim Kent, Eric Green, and Webb Miller

A statistical estimate based on a simple measure of conservation between short orthologous segments in the human and mouse genomes suggests that about 5% of the human genome may be under purifying selection, but does not definitively tell us which segments are under selection. To discover these, we will soon be able to compare the entire reference human genome to many other reference vertebrate genomes (and perhaps to even more distantly related genomes). The bioinformatic grand challenge here is to use this data to trace back as far as possible the evolutionary history of each base in the reference human genome. Then, by analyzing the detailed patterns of molecular evolution of segments of the genome over a long period, we might hope to recognize the signatures of purifying selection in different kinds of functional elements, and also detect instances of positive selection leading to new functions. We are trying to address this grand challenge through the development of refined sequence alignment methods and new models for molecular evolution. These models include context dependencies between substitutions at adjacent bases, e.g. as in the process of CpG decay, and that allow for different rates of evolution in different parts of the chromosomes, as has been confirmed in recent analysis. Gene-finding hidden Markov models for DNA have been extended to process multiple alignments instead of single sequences, and to include context-dependent models of evolution. Finally, a method of assigning p-values to conserved elements based on models of molecular evolution has been developed and applied to predict functional elements from a multiple alignment of the NISC Comparative Sequencing Program data in the human CFTR region. While these methods are still in early development and the results from them preliminary, we are encouraged about the potential of this line of research, if a bit daunted by the complexity of the problem.