University of Illinois Urbana-Champaign

Machine learning for complex biological systems at multiple scales

Nambiar, Ananthan

This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.

Permalink

https://hdl.handle.net/2142/130075

Description

Title
Machine learning for complex biological systems at multiple scales
Author(s)
Nambiar, Ananthan
Issue Date
2025-06-09
Director of Research (if dissertation) or Advisor (if thesis)
Maslov, Sergei
Doctoral Committee Chair(s)
Maslov, Sergei
Committee Member(s)
  • Anastasio, Mark
  • Milenkovik, Olgica
  • Lam, Fan
Department of Study
Bioengineering
Discipline
Bioengineering
Degree Granting Institution
University of Illinois Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
  • Machine Learning
  • Complex Systems
  • Biological Systems
  • Systems Biology
  • Computational Biology
  • Proteins
  • Genes
  • Liver Disease
Abstract
Understanding biological systems requires modeling the complex, nonlinear interactions among their components. While high-throughput technologies have made vast biological datasets available, interpreting these data remains a challenge. In this thesis, I present machine learning approaches that tackle this problem across multiple biological scales—ranging from protein function, to gene regulation, to genome-wide association studies. At the molecular scale, I develop a Transformer-based protein language model to generate task-agnostic sequence representations. These representations are finetuned for multiple downstream tasks, including protein family classification, protein-protein interaction prediction, and disordered region annotation. The model outperforms or matches state-of-the-art task-specific methods while maintaining generality. At the regulatory scale, I introduce FUN-PROSE, a deep learning framework for predicting condition-specific gene expression in fungi. By integrating promoter sequences with transcription factor expression data, FUN-PROSE captures complex regulatory logic and achieves high accuracy across multiple fungal species. Interpretation of the model reveals biologically relevant sequence motifs and transcription factor–gene interactions, offering insights into gene regulation. At the level of complex trait genetics, I develop a machine learning–guided GWAS framework to identify genetic variants associated with metabolic dysfunction-associated steatotic liver disease (MASLD) in individuals with obesity. By training a machine learning model on clinical and biochemical features to generate a continuous MASLD risk score (the I-MASLD score), I use this quantitative phenotype in a genome-wide association study. This approach recovers known MASLD-associated genes such as PNPLA3 and PTEN, while also implicating novel loci including HERC2 and GRIA3. Together, these models demonstrate how machine learning can leverage patterns in biological data to generate interpretable, predictive models of complex biological systems.
Graduation Semester
2025-08
Type of Resource
Thesis
Handle URL
https://hdl.handle.net/2142/130075
Copyright and License Information
Copyright 2025 Ananthan Nambiar

Owning Collections

Graduate Dissertations and Theses at Illinois PRIMARY
Graduate Theses and Dissertations at Illinois