Improving, validating, and interpreting amino acid substitution models

Proteins drive much of the chemistry of life. They control many of the reactions that allow organisms to survive, and they are essential structural elements that determine life’s form. All proteins are long chains of subparts called Amino Acids of which there are 20 common types. Substitution of one amino acid for another may have either a small or a large effect on the form and function of a protein. Knowing the probability of substituting one amino acid can help determine the evolutionary history of a protein. This research will improve our estimates of the substitution probabilities between the common amino acid. This information will then improve software that is essential for reconstructing the tree of life. The project will train multiple students in statistics and data science. In addition, outreach activities will improve science literacy by promoting critical use and editing of online information sources. This research will improve the inference of amino acid substitution models by excluding likely alignment errors from training data. Analysis of taxon-specific differences in protein evolution should then be revealed by the removal of universal artifacts. Training of amino acid substitution models on data specific to different types of protein structural characteristics will also improve understanding of substitution probabilities by uncovering how different structural types of proteins evolve. The resulting structure-informed mixture and/or partition models can then be used to improve phylogenetic inference. The methods develop will allow for changes in amino acid frequency over time. These methods will be validated by comparison of predictions and experimental effects. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.