Bridging AI and biological discovery: novel applications of language models in enzyme prediction and protein classification = AI와 생물학적 발견의 연결: 효소 예측 및 단백질 분류에서 언어 기반 딥러닝 모델의 적용

목차 (Table of Contents)

• Content
• Abstract 3
• Chapter 1. Introduction 6
• A. Bioinformatics with Biological Big Data and AI Advances 6
• B. An Overview of Artificial Intelligence 7
• C. Machine Learning and Deep Learning in Biology 11
• D. Application of Deep Learning in this Research 15
• Chapter 2. Prediction Model for Specific Enzyme Reaction: SPRINT
• (Substrate-Product-Reaction Inference with deep Neural neT) 17
• A. Background 17
• 1) Overview of Enzyme Reaction Screening 17
• 2) Prior Research 19
• 3) Objective 20
• B. Material and Methods 21
• 1) Data Collection from Three Reaction Databases 21
• 2) Data Curation and Preprocessing for SPRINT 22
• 3) Model Architecture and Training Procedures 24
• 4) Data for Case Study: Plastic DB 26
• 5) Methodologies for Validation and Evaluation 28
• C. Results 29
• 1) Predictive Performance of SPRINT 29
• 2) Model Performance by Enzyme EC Classification 30
• 3) Evaluation Across Protein Sequence Identity 31
• 4) External data validation: Plastic DB (PETase) 32
• Chapter 3. Prediction Model for Enzyme Turnover Number: KITCAT
• (Kinetic Inference Tool for Catalysis Activity Turnover rate) 35
• A. Background 35
• 1) Description of kcat Parameter in Enzyme Kinetics 35
• 2) Prior Research 36
• 3) Objective 37
• B. Material and Methods 38
• 1) Collection of kcat Dataset 38
• 2) Data Preprocessing for KITCAT 41
• 3) Model Structure and Training Methods 41
• 4) Methodologies for Validation and Evaluation 44
• C. Results 46
• 1) Predictive Performance of KITCAT 46
• 2) Model Performance by Enzyme EC Classification 47
• 3) Evaluation Across Protein Sequence Identity 49
• 4) Model Comparison with Another Research 49
• Chapter 4. Concluding Remarks and Further Works 51
• A. Discussion on the Application of Biological Data Using Deep
• Learning 51
• B. Conclusion 54
• References 56
• 국문초록 64