15 Feb 2013
Sean Eddy
HHMI, Janelia Farms
Sequence database homology searching is one of the most important applications in computational molecular biology. There are undoubtedly many homology relationships that current computational sequence analysis methods are unable to resolve. The frontier in this field is to increase our resolution by building increasingly realistic, complex, parameter-rich models for sequence comparison, using probabilistic inference methods including hidden Markov models (HMMs), leveraging the massive amount of available comparative sequence data. I will discuss how the homology search problem is formally a problem of predicting the probabilities of DNA and protein sequence evolution on large time scales. I will talk about how the historically distinct fields of sequence homology searching and phylogenetic inference are starting to merge, at long last.