 |
Center for Nonlinear Studies |
Due to the explosion of data in the past decade, a great deal of focus has shifted from the understanding of single molecules to the synthesis of information about a multiplicity of related molecules with the intent to rationally design new molecules based on machine learning or similar concepts. In this talk, I will discuss my work in two disparate areas, linked primarily by this shared ethos: first, I will discuss antibiotic design for Gram negative bacteria and next I will discuss conotoxin design for therapeutic applications. My discussion of antibiotic design will focus on preliminary work and proof-of-concept of a new method for identifying a library of fragments for the creation of antibiotic hybrids that are able to avoid bacterial multidrug resistance. My discussion of conotoxins will focus on our recent work in using a rational network-based approach to identify key structures for homology modeling that will require the least amount of experimental effort to construct a library of structures of extant conotoxin sequences. I will also discuss ideas for experiments to probe kinetic and thermodynamic control of conotoxin structural ensembles, beyond characterization of their native structures. Overall, this work explores proof-of-concept for several novel techniques as well as laying out important questions in the field of rational molecular design.
Host: David Métivier