Abstract by Danae Stephens
Physics and Astronomy
Dennis Della Corte
Developing a reliable initial model for protein folding predictions.
High-quality protein structures are needed for a variety of scientific pursuits such as medical research. The current drawback is that most proteins cannot be crystallized through traditional experimental methods. Due to this computation predictions have come to the forefront as we move forward with protien predictions. Our lab utilizes a Convoluted Nural Network (CNN) to generate structural constraints for proteins; primarily focusing on molecular distances with auxiliary predictions for torsion angles and secondary structures. My piece of this was to covert these predictions into a full atomistic protein structure. Once completed this pipeline will allow accurate predictions of any amino acid sequence of interest, thus allowing scientists in multiple fields to quickly generate and asses protein structures for reserch.