“Predicting Function from Sequence”
In my graduate and postdoctoral research, I applied various statistical methods to predict functional information about biological molecules ranging from the large macromolecular machine, the ribosome, to small molecule natural products. Using Statistical Coupling Analysis (SCA), I predicted networks of coevolving bases in the large subunit rRNA. The functional importance of bases in these networks was confirmed using a continuous culture assay with deep sequencing. I developed a machine learning-based bioinformatics tool that predicts natural product bioactivity using the sequence of the natural product’s biosynthetic gene cluster. For this initial study I focused on different antimicrobial activities. The resulting machine learning classifiers can predict natural product activity with accuracies as high as 79%. I also determined that some biosynthetic genes are highly associated with certain activities, linking molecular features to bioactivity.
~ Coffee/tea will be served prior to lecture~