Application of Machine Learning to Study Characteristics of Heat shock Protein 104 (HSP 104)
Main Article Content
Abstract
Record ID: 275
Award(s): Excellence in Undergraduate Research Mentorship; Excellence in Research Communication
Program Affiliation: Capstone
Presentation Type: Poster
Abstract: Newly produced proteins or already folded proteins may get damaged and lose their 3D structures that is crucial for their biological function. This can happen because of heat shock, or stress, leading to neurodegeneration and various other diseases. The Heat Shock protein 104 AAA+ is a chaperone protein, so called for its duty to prevent unwanted conformations of proteins, mediates protein quality control via protein degradation or disaggregation. Proteins belonging to the AAA+ family (ATPases associated with diverse cellular activities) play a crucial role in preserving protein stability. The Hsp104 from Saccharomyces cerevisiae (brewers yeast) are members of the AAA+ family that support protein quality control by unfolding aberrant or toxic proteins. Though we have identified the purpose of this enzyme, how this protein performs disaggregation is unclear. In this study we implement a clustering method on molecular dynamics simulations. After having simulated the action of the Hsp104 on its substrate and having obtained the timeline of how the shape of Hsp104 changes, we identified and picked out most representative 3D molecular shapes that are important for understanding the behavior of the system. After identifying biochemically relevant descriptors based on the literature, we apply machine learning algorithms to identify the structural features that best describe the identified conformations. Results from this study can help advance research on treatment for neurodegenerative diseases which proceed by accumulation and aggregation of misfolded proteins.