Graduation Semester and Year
Spring 2026
Language
English
Document Type
Dissertation
Degree Name
Doctor of Philosophy in Computer Science
Department
Computer Science and Engineering
First Advisor
Jean X. Gao
Second Advisor
Junzhou Huang
Third Advisor
Qilian Liang
Fourth Advisor
Dajiang Zhu
Abstract
Functional genomics seeks to understand how genes and regulatory elements orchestrate biological processes and adapt under diverse conditions. Two critical challenges in this field are deciphering the regulatory roles of long noncoding RNAs (lncRNAs) and modeling cellular transcriptional dynamics under therapeutic perturbations. While lncRNAs are increasingly recognized as key regulators in health and disease, their functions remain incompletely characterized. Concurrently, advances in single-cell transcriptomics reveal how gene and pathway activities rewire across drug dosage gradients, highlighting the need for computational frameworks that are predictive, interpretable, and capable of integrating molecular interactions with dynamic cellular states. This dissertation develops a unified research program of attention-based deep learning frameworks for functional genomics that progressively scale across three levels of biological organization: pairwise molecular interactions, cellular adaptation under perturbation, and transcriptome-wide functional annotation. At the molecular level, we propose an attention-based model for predicting lncRNA–miRNA interactions by integrating sequence and structural information, achieving strong predictive performance with interpretable attention maps. Building on these attention-driven principles, we extend to the cellular level with a hierarchical graph-based framework that models dose-specifc transcriptional dynamics at single-cell resolution. By constructing pathway–gene–cell graphs augmented with dosage embeddings, the model captures resistance-associated rewiring, quantifies intra-dose heterogeneity, and generalizes to unseen drug concentrations. Extending these methodological foundations to the transcriptome scale, we present FALCON (Functional Annotation of LncRNA via Cross-layer Attention Network), a graph-based framework for uncertainty-aware functional annotation of human lncRNAs. FALCON integrates six layers of biological evidence into a heterogeneous knowledge graph and employs cross-layer attention to learn unified lncRNA representations, while Monte Carlo Dropout provides calibrated uncertainty that reflects functional coherence rather than annotation density. Together, these contributions establish scalable, interpretable, and uncertainty-aware methods that bridge molecular interaction modeling, cellular adaptation, and integrative functional annotation of noncoding RNAs.
Keywords
lncRNA, miRNA, deep learning, graph neural networks, functional genomics, single-cell transcriptomics, attention mechanisms, bioinformatics, RNA interaction prediction, heterogeneous graphs
Disciplines
Other Computer Engineering | Other Electrical and Computer Engineering
License
This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivative Works 4.0 International License.
Recommended Citation
Satyal, Sachit, "Deep Learning Frameworks for Functional Genomics: lncRNA–miRNA Interactions, Dose-Specific Graph Modeling, and Complete Graph-Based lncRNA Functional Annotation Using Cross-Layer Attention" (2026). Computer Science and Engineering Dissertations. 2.
https://mavmatrix.uta.edu/cse_dissertations2/2