Date Approved
4-4-2025
Embargo Period
4-4-2025
Document Type
Thesis
Degree Name
Master of Science (M.S.)
Department
Chemistry and Biochemistry
College
College of Science & Mathematics
Advisor
Hao Zhu, Ph.D
Committee Member 1
Zhu Hao, Ph.D.
Committee Member 2
Thomas Keck, Ph.D.
Committee Member 3
Zhihong Wang, Ph.D.
Abstract
Massive chemical data is now publicly available through publications and various online databases. However, most online databases only provide data unsuitable for modeling purposes and limited machine learning (ML) modeling for those data. As a public chemical toxicity evaluation tool to resolve these issues, the Toxicology Universe (ToxiVerse) portal hosts curated chemical structures and toxicological/pharmaceutical data of 26,000 compounds which can be used directly for modeling purposes. Furthermore, users can upload chemical datasets to extract all involved bioassays for compounds inside as biological profiles. ToxiVerse also features automatic QSAR modeling of uploaded datasets using multiple chemical descriptors and algorithms provided by the ToxiVerse portal synchronously and predicts target chemicals for their toxicity potentials and fundamental pharmaceutical properties. A knowledge-based deep neural network (k-DNN) is also integrated to identify potential estrogenic toxicants and their toxicity pathways. The ToxiVerse portal provides the scientific communities with a robust online public platform to obtain machine-readable datasets and perform modeling for desired toxicity and pharmaceutical properties.
Recommended Citation
Shen, Yitao, "TOXIVERSE: AN ONLINE COMPREHENSIVE CHEMICAL PROFILING, MODELING, AND DATA ANALYSIS PLATFORM" (2025). Theses and Dissertations. 3343.
https://rdw.rowan.edu/etd/3343
