Epiphany2: A Novel CNN-Transformer Method for Predicting 3D Chromatin Structure from Epigenetic Data 公开
Belov, Alex (Spring 2025)
Abstract
Understanding the three-dimensional (3D) organization of chromatin and its regulation by the epigenome is critical for unraveling the complexities of gene expression, cellular differentiation, and disease mechanisms. While current models leveraging 1D epigenomic data to predict 3D chromatin structure have shown promise, many suffer from key weaknesses, including a limited ability to capture cell-type-specific interactions and a lack of global contextual understanding of chromatin dynamics. This thesis presents a novel model architecture with state-of-the-art performance in predicting Hi-C contact maps. Our model is the first to use transformer layers to capture long-range dependencies for this task.
Table of Contents
Chapter 1 (8)
Introduction (8)
Chapter 2 (11)
2.1 Background (11)
2.2 Motivation (13)
2.3 Challenges (14)
Chapter 3: Existing Works (16)
3.1 Epigenome → Hi-C (17)
3.2 Similar Biological Tasks (20)
Chapter 4: Technical Designs (22)
Chapter 5: Experimental Settings (25)
5.1 Preprocessing (25)
Chapter 6: Experiments & Results (27)
Chapter 7: Conclusion (30)
Chapter 8: Limitations & Future Work (31)
Bibliography (33)
About this Honors Thesis
| School | |
|---|---|
| Department | |
| Degree | |
| Submission | |
| Language |
|
| Research Field | |
| 关键词 | |
| Committee Chair / Thesis Advisor | |
| Committee Members |
Primary PDF
| Thumbnail | Title | Date Uploaded | Actions |
|---|---|---|---|
|
|
Epiphany2: A Novel CNN-Transformer Method for Predicting 3D Chromatin Structure from Epigenetic Data () | 2025-04-09 10:58:19 -0400 |
|
Supplemental Files
| Thumbnail | Title | Date Uploaded | Actions |
|---|