Protein Structure Evolution and
Sequence alignment methods are common for comparisons of both DNA and protein sequences. In the case of proteins, additional information is available in the spatial structure of the protein. Alignment algorithms which make use of structural information have been developed, but lack formal modeling assumptions. Dr. Schmidler and I have developed the first fully probabilistic evolutionary model which incorporates both sequence and structure. The paper is now available via advance access from Molecular Biology and Evolution.
Particle Learning for Probabilistic Deterministic Finite Automata
Over the summer I worked with Ryan Prenger and Dan Merl of Lawrence Livermore National Laboratory to develop sequential learning algorithms for PDFA, which can provide an attractive alternative to HMMs in many settings due to reduced computational requirements. An R package which implements our methods is pending the labortorary's information release process.