Jianyuan Deng

I am a PhD student in the Department of Biomedical Informatics at Stony Brook University, under supervision of Dr. Fusheng Wang. I also work with Dr. Dimitris Samaras and Dr. Iwao Ojima.

I received my M.Phil. degree from School of Pharmacy at The Chinese University of Hong Kong. Before that, I obtained my B.Sc. degree in Pharmaceutical Sciences from Sun Yat-Sen University.

Email  /  CV  /  Google Scholar  /  Github

My research interests broadly include informatics, machine learning, drug discovery, and oral drug absorption. Currently, my work primarily focuses on molecular property prediction & molecule generation with deep learning.

We conduct a systematic study on molecular property prediction and reflect on multiple key elements. We also highlight that activity cliffs can significantly impact prediction. Furthermore, we explore into potential causes why representation learning models can fail and emphasize that dataset size is essential for representation learning models to excel. (supplement; repository).

We conduct a comprehensive survey on AI-driven drug discovery, mainly on molecular property prediction and molecule generation. We also release a Github repository (link; Initial Release Date: 2021-06-06) for a collection of related papers in AI-driven Drug Discovery.

We conduct a retrospective cross-sectional study on opioid overdose by mining a large-scale electronic health records database and also explore its association with opioids' properties.

We present an LSTM model for predicting opioid use disorder among patients taking opioid medications based on their electronic health records.

We present an LSTM model for predicting opioid overdose among patients with opioid prescriptions based on their electronic health records.

We develop a deep reinforcement learning framework to discover potential lead compounds as better opioid antagonists with enhanced brain retention ability.

We propose an informatics approach to identify the most influential carriers, transporters, enzymes and targets underlying drug-drug interactions.

We present an in silico drug absorption tract, ISDAT, to simulate human oral drug absorption of clonazepam and midazolam.

We put forth and implement a framework for high-capacity color QR codes equipped with our methods, called HiQ. A fast color QR code decoding algorithm is also presented.

We conduct a comprehensive review on food effect and examine how it affects oral absorption extent and rate based on the Biopharmaceutics Drug Disposition Classification System.

We present an agent-based model to represent baicalein's pharmacokinetic properties based on a bi-directional transport study in the Caco-2 cell monolayer.

Design and code stolen from Jon Barron.