Exploring the entire Olfactory Receptor agonism space through AI-driven pocket analyses
- 주제(키워드) Olfactory Receptors , Template-based modeling , AlphaFold2 , Structures-based information , Agonists , Property Prediction , Deorphanization
- 발행기관 서강대학교 일반대학원
- 지도교수 성봉준
- 발행년도 2026
- 학위수여년월 2026. 2
- 학위명 석사
- 학과 및 전공 일반대학원 화학과
- 실제URI http://www.dcollection.net/handler/sogang/000000082481
- UCI I804:11029-000000082481
- 본문언어 영어
- 저작권 논문은 저작권에 의해 보호받습니다.
목차
1. Abstract 5
2. Introduction 6
3. Results 10
3.1. Overview of the research 10
3.2. Template-based approaches can improve the accuracy of active-state OR modeling 11
3.3. Prediction of agonist properties using active structures 16
3.4. Active structures improve the prediction of agonist prediction 25
4. Discussion 28
4.1. Active state template structures improve the quality of OR structural model 28
4.2. Structural information provides better clues for agonists than sequence-based features 29
4.3. Limitations of our work 29
4.4. Future works 30
5. Conclusion 31
6. Methods 32
6.1. Dataset Preparation 32
6.2. Active Structure Modeling 34
6.3. Agonist and Property Prediction with modeled active structures 36
6.3.1. Architecture 36
6.3.2. Preprocessing 39
6.3.3. Cross Attention 40
6.3.4. Loss function 43
6.3.5. Ligand Embedding 44
6.3.6. Protein language model for OR representation 45
6.3.7. Property Prediction 45
6.3.8. Training condition 47
7. References 48
8. Abstract (English) 52
