Data Mining in Exploring Protein Thermodynamics and Kinetics Relationships

• Activity Number: 25
• Type: Contributed
• Date/Time: Sunday, July 31, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Amy Wagaman*+ and Sheila Jaswal
• Companies: Amherst College and Amherst College
• Keywords: data mining ; protein folding ; classification ; regression ; Isoband ; covariance estimation

Abstract

Information on protein stability, folding and unfolding kinetics is critical to understanding the normal biological function of many proteins, as well as the misfolding and aggregation properties of a growing number of proteins found to be involved in neurodegenerative and other diseases of conformation. We assembled a database to investigate relationships between energetic quantities related to protein folding thermodynamics and kinetics (beyond the known deterministic relationships) and protein structure and function. In this article, we tackle a classification problem and show improvements on existing techniques. We also extend folding relationships to unfolding relationships in a related prediction problem, while trying to use more elementary predictors which do not require structural information for the protein beyond its amino acid composition. Through this and related work, we investigate the existence of classes of protein folding landscapes that can be distinguished solely on the basis of structural properties.