This research focuses on building a computational understanding of the allosteric effect behind GPX4 protein mutations and exploring the drug design possibility as a potential approach to tackle Alzheimer’s Disease. I used high-performance computing clusters to perform molecular dynamics simulation on GPX4 protein variants Applied K-means clustering, MD-Markov State Model, and other machine learning algorithms to analyze structural differences between GPX4 wild type protein system and that of its mutant variants. Using the energetic networks of protein structures, I proposed a novel approach of examining protein allosteric effects via the shortest electrostatic pathways.

This research focuses on devising generalized methods to extract characteristics of dynamic graphs, which support Q-value estimation of reinforcement learning based routing algorithms in multi-hop wireless mobile networks. Using Tensorflow, I implemented deep recurrent (long short-term memory) neural networks to estimate meeting intervals between pairs source and destination nodes in mobile networks with decentralized local information.

This research focuses on creating usable privacy configuration schemes which allow active user privacy choices. We are developing a chromium-based GPC) signals in the outgoing HTTP requests, allowing users to opt-out from networks sharing and selling personal information for ad purposes. We implemented the software architecture to assign users to different privacy schemes for A/B testing automatically. By collecting user behavioral data during their interactions with the extension, we perform usability studies of various hypothetical privacy permission schemes implemented at the browser level.