Machine learning for large and small data biomedical discovery

In modern biomedicine, the role of computation becomes more crucial in light of the ever-increasing growth of biological data, which requires effective computational methods to integrate them in a meaningful way and unveil previously undiscovered biological insights. In this dissertation, we introduce a series of machine learning algorithms for biomedical discovery. Focused on protein functions in the context of system biology, these machine learning algorithms learn representations of protein sequences, structures, and networks in both the small- and large-data scenarios. First, we present a deep learning model that learns evolutionary contexts integrated representations of protein sequence and assists to discover protein variants with enhanced functions in protein engineering. Second, we describe a geometric deep learning model that learns representations of protein and compound structures to inform the prediction of protein-compound binding affinity. Third, we introduce a machine learning algorithm to integrate heterogeneous networks by learning compact network representations and to achieve drug repurposing by predicting novel drug-target interaction. We also present new scientific discoveries enabled by these machine learning algorithms. Taken together, this dissertation demonstrates the potential of machine learning to address the small- and large-data challenges of biomedical data and transform data into actionable insights and new discoveries.

Name thesis:degree_name

Ph.D.

Level thesis:degree_level

Dissertation

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2022

Author dc:creator

• Luo, Yunan

Contributors dc:contributor

• Peng, Jian
• El-Kebir, Mohammed
• Han, Jiawei
• Ma, Jianzhu
• Cho, Hyunghoon

• Computer science

Statement dc:rights

• Copyright 2021 Yunan Luo

Language dc:language

en, eng

Handle dc:identifier

http://hdl.handle.net/2142/113873

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F113873