(6cr) Integrating Computational and Experimental Methods to Discover Disease Causes and Design Protein Therapeutics

With the advent of genetic sequencing, there has been a clear trend in medicine to develop increasingly specialized products. Antibody therapeutics effective for only a specific gene mutation in cancers and individualized genome sequences are just two examples. As more data becomes available, there is a clear need for experimental and computational methods that can identify the most significant facts and utilize them to generate effective results. My vision is to have a combined computational and experimental group that identifies the causes of diseases and develops treatments for them. The graduate and postdoctoral research I have completed has created a strong foundation for achieving this goal.

At Penn State University, my graduate research was on the development of computational tools for protein engineering and de novo design, with a specific focus on antibodies. The Optimal Complementarity Determining Regions method (OptCDR) was developed to design the antigen-binding portions of antibodies, known as CDRs. Using Mixed Integer Linear Programming optimization, a rotamer library, and molecular mechanics force fields, libraries of any user-specified size of antibody CDRs against any specified antigen epitope can be generated. Other work focused on creating a database of modular antibody parts that can be accurately assembled to model antibody structures and using these parts in the Optimal Method of Antibody Variable region Engineering (OptMAVEn) to de novo design fully human antibodies. As computation speeds increase and force fields are refined, novel methods such as OptCDR and OptMAVEn will be essential tools to design protein therapeutics for specific objectives.

The postdoctoral research I have conducted at the University of California, Santa Barbara is experimental work to find environmental factors associated with diseases. Through a combination of magnetic selection and fluorescence activated cell sorting of a large peptide library and next-generation sequencing, it is possible to identify a large database of peptides bound by a patient's antibodies. In turn, computational analysis of the identified amino acid sequences from multiple disease and control patients can find motifs that are disease specific and cross-reactive. These motifs can be used to hypothesize the environmental factors that are contributing to or causing the disease.

As experimental methods improve, increasingly vast databases will become available. It is imperative that effective computational methods are available to analyze and utilize this data. However, computations alone are insufficient and they must be coupled with experimental techniques to evaluate the generated hypotheses. My future plans are to build upon what I have done and continue to develop computational and experimental methods for identifying causes of diseases and developing treatments for them. The computational and experimental research I have done so far is an excellent foundation for developing integrated methods to utilize the large amounts of data that are becoming available.