The alignment of protein structures is essential for the study of protein function. Superposing the binding cavities of one protein onto another can reveal shape similarities that explain why two proteins bind a similar molecular partner. Superposition can also expose critical differences that explain why proteins sometimes bind very different partners. These observations, made by experts in structural biology with the computational aid of protein structure alignments, help construct our atomic-level understanding of protein binding preferences. Protein binding preferences lie at the crux of many pressing challenges in protein engineering, systems biology, drug design, and other fields.
This project considers a new approach to protein structure alignment that generates superposition by maximizing the overlapping volume between binding cavities. Two core research directions will be considered: (1) The development of specialized optimization techniques that rapidly and reliably explore the continuous space of alignments and (2) the usage of biological and biophysical data to enhance the alignment process.
Intellectual Merit: The proposed work advances the study of protein structure comparison with new representations of protein structure and new geometric superposition techniques. A novel focus on closely related proteins also diverges from existing methods, which generally target proteins with distant evolutionary ties. The project advances techniques in mathematical optimization for accommodating the noisy and expensive estimation of overlapping cavity volume. Accommodations of this nature are an emerging challenge in optimization, where the function being optimized is frequently the result of imprecise and expensive computations, like simulations. The low dimensionality of the cavity alignment problem, coupled with the opportunity to optimize in the presence of noise and imprecision creates a unique opportunity to advance techniques for this general challenge in optimization. In structural biology, the proposed work will add automation and precision to existing workflows, enabling the examination of more structures than humans can fully consider and identifying molecular subtleties that might otherwise be overlooked.
Broader Impacts: The project also focuses on developing interactive software for protein structure comparison. This software will provide a visual and intuitive interface, helping students and scientists from non-computational background to better use and understand software in structural bioinformatics. The research itself and the development of the interface software will expose graduate and undergraduate students with mono-disciplinary backgrounds to an interdisciplinary research program in bioinformatics, and train them in cross-disciplinary communication.
