- Zixuan Cang (Mathematics, NCSU)
- Jingjie Hu (Mechanical Engineering, NCSU)
Prerequisites: Basic algebraic topology, linear algebra, basic programming.
Outline: Nanoparticles with specifically designed coating can interact with target cancer cells that overexpress certain membrane proteins. The understanding of adhesive interaction at the nanoscale between functionalized nanoparticles and biological cells is of great importance to develop effective theranostic nanocarriers for targeted cancer therapy . Topological data analysis is an emerging field that utilizes algebraic topology to describe structures of complex data . It has led to competitive tools for drug design . In this project, we will develop topological data analysis-based methods to characterize the molecular structures in protein interactions and link the structures to functions.
Objectives: Examining molecular structures of protein-protein interactions from molecular dynamics simulations of nanoparticle-cancer cell interactions with topological data analysis; Identifying topological characteristics and differences for different nanoparticle coatings. The topological summary and the experimentally measured adhesion strength will be used to build machine learning models to infer the structure-function relationships.
Outcomes: A concise topological representation of the complex protein-protein interactions that arise in nanoparticle-cancer cell interactions and a machine learning model that predicts adhesion strength from the topological representation. Results will provide a tool for high-throughput computational prediction of adhesion strengths which can help the design of nanoparticle coating.
References: Hu, J., et al., Investigation of adhesive interactions in the specific targeting of Triptorelin-conjugated PEG-coated magnetite nanoparticles to breast cancer cells. Acta Biomater, 2018. 71: p. 363-378.  Wasserman, L., Topological data analysis. Annu Rev Stat Its Appl, 2018. 5: p. 501-532.  Cang, Z. and G.-W. Wei, Topology based deep convolutional and multi-task neural networks for biomolecular property predictions. PLoS Comput Biol, 2017. 13: p. e1005690.