|
Abstract:
|
Identification and evaluation of structural similarity between proteins is a widely studied problem in phylogenetics and molecular biochemistry. A previously introduced statistical model for structural evolution of proteins significantly improved phylogenetic inferences that are based only on the amino acid sequences of proteins. We extend this protein structure evolution model to incorporate spatial dependence among neighboring amino acid positions. The result is a multivariate diffusion process convolved with a spatial birth-death process which comes with little additional computational cost or analytical complexity compared to the site-independent model. Via simulation studies, we show this extended model generates more realistic protein structures, including insertions, and demonstrate improvement of structure alignments, including alignment uncertainty and estimation of evolutionary distance.
|
