A HYBRID DEEP-MACHINE LEARNING APPROACH FOR PREDICTING ENVIRONMENTALLY RESPONSIVE TRANSGENERATIONAL DIFFERENTIAL DNA METHYLATED REGIONS (EPIMUTATIONS) IN THE GENOME - research dissertation

Aug 5, 2026

Machine Learning and Deep Learning are central to predictive modeling, yet selecting informative features often requires domain expertise and deep networks demand large data and extensive hyperparameter tuning. This dissertation proposes a hybrid model that combines a deep learning component for feature extraction with a machine learning component for final prediction. The approach uses XGBoost to improve performance on imbalanced data and reduces tuning effort. It also enables visualization of sequence motifs and ranking of feature importance. Systematic analysis identifies optimal network size and extraction layers.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages207

Reading Time8 h 42 min

AccessFree to download

Acknowledgment

Abstract

List of Tables

List of Figures

Chapter One: Introduction

Problem Definition

Contribution

Chapter Two: Related Works

Representation Learning

Hybrid Approaches

The application of hybrid models in epigenetics

Chapter Three: Method

A HYBRID DEEP-MACHINE LEARNING APPROACH FOR PREDICTING ENVIRONMENTALLY RESPONSIVE TRANSGENERATIONAL DIFFERENTIAL DNA METHYLATED REGIONS (EPIMUTATIONS) IN THE GENOME By PEGAH MAVAIE A dissertation submitted in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY WASHINGTON STATE UNIVERSITY School of Electrical Engineering and Computer Science MAY 2023 © Copyright by PEGAH MAVAIE, 2023 All Rights Reserved © Copyright by PEGAH MAVAIE, 2023 All Rights Reserved To the Faculty of Washington State University: The members of the Committee appointed to examine the dissertation of PEGAHMAVAIE find it satisfactory and recommend that it be accepted. Lawrence B. Holder, Ph.D., Chair Michael K. Skinner, Ph.D. Ananth Kalyanaraman, Ph.D. ACKNOWLEDGMENT First and foremost, I would like to thank my advisor, Dr. Lawrence Holder, for his invaluable guidance, support, and encouragement throughout my Ph.D. studies. His expertise and insight have been crucial to my academic growth and personal development, and I am deeply grateful for the time and effort he has invested in me. I would also like to thank Dr. Michael Skinner for introducing me to this exciting area of research. I am grateful for his encouragement and support throughout this study. I would like to express my appreciation to my committee member, Dr. Ananth Kalyanaraman, for his advice and feedback in shaping my dissertation. His critical evaluation and constructive feedback have helped me improve the quality of my work. I would like to express my profound gratitude to the Templeton Foundation for providing the essential funding that enable this research. I am also grateful to my colleagues, friends, and family, for their support and encouragement during this challenging journey. Their love, understanding, and encouragement have been a source of strength and inspiration to me. In conclusion, I am grateful to everyone who has played a part in my dissertation. Your support, guidance, and encouragement have been invaluable, and I will cherish these memories for years to come. A HYBRID DEEP-MACHINE LEARNING APPROACH F0R PREDICTING ENVIRONMENTALLY RESPONSIVE TRANSGENERATIONAL DIFFERENTIAL DNA METHYLATED REGIONS (EPIMUTATIONS) IN THE GENOME Abstract by Pegah Mavaie, Ph.D. Washington State University May 2023 Chair: Lawrence B. Holder With the progress of Machine Learning (ML) in the past few decades, ML has become a prominent solution for different applications. Representing data with the most informative set of features is essential for learning accurate models. In many datasets, this process can be labor-intensive and requires the user to have enough background knowledge about the domain to select relevant features. Deep learning (DL) as a part of machine learning has improved the predictive model performance since the early 2000s by automatically extracting, analyzing, and understanding useful information directly from the raw data. While DL allows automatic feature extraction from raw data, it requires a large amount of data and significant hyperparameter tuning. To overcome these challenges, we propose a hybrid model. In the hybrid model a DL component is used to extract features from the data and the ML component is used for the final prediction. The model uses XGBoost to deal more effectively with imbalanced data using the boosting process and reduce the need for hyper-parameter tuning. Additionally, the hybrid model allows the visualization of sequence motifs corresponding to the extracted features and the ranking of the importance of these features for the prediction task. A systematic analysis of the hybrid model is performed to identify the best settings in terms of the size of the deep network and the layer from which to extract features. The hybrid model was developed in the context of a particular target application, i.e., the identification of regions in the genome with susceptibi

FAQs

What main problem does the dissertation address in predictive modeling?

Accurate learning depends on informative feature representation, but manual feature selection can be labor-intensive and domain-dependent. Deep learning can automate feature extraction but typically requires large datasets and heavy hyperparameter tuning.

How does the proposed hybrid model work?

The model uses a deep learning component to extract features from input data, then applies an ML component to produce final predictions. XGBoost is used to handle imbalanced data more effectively and reduce the need for hyperparameter tuning.

What biological application and evaluation results are reported?

The model targets identification of genomic regions associated with susceptibility to differential DNA methylated regions (DMRs) caused by environmental toxicant exposure. Results indicate the hybrid model outperforms deep learning alone and machine learning alone for DMR epimutation prediction, and it can identify exposure-specific and disease-specific transgenerational DMR epimutations.

Aug 5, 2026

Upload by Lucas Martin