- Title
- The ligand docking experiments between olfactory receptors and allergen molecules and the classification of olfactory receptors using machine learning methods
- Creator
- Wang, Pu
- Relation
- University of Newcastle Research Higher Degree Thesis
- Resource Type
- thesis
- Date
- 2022
- Description
- Research Doctorate - Doctor of Philosophy (PhD)
- Description
- Allergens in perfumes potentially cause allergic reactions in humans. Multiple ectopic expressions of olfactory receptors (ORs) recognize odors, and this provides the possibility that potential allergen components in perfumes bind specifically to the corresponding ORs and cause allergic reactions. The lack of known high-resolution images of the structures of ORs has largely limited the relevant research. As a widely used method to predict protein structure, homology modeling still has some limitations in being applied to predict the structure of ORs due to the dissimilarity between ORs and G-protein-coupled receptors (GPCRs). This research applied ligand docking experiments to determine the interactions of ORs with their corresponding ligands and explore the potential binding sites. For these experiments, based on the limitations of traditional homology modeling methods for OR-related studies and the problem of ligand docking between potential allergen components in perfumes and the corresponding ORs, the following solutions were proposed: 1. Sequence alignment dataset based on structural features of ORs We built the ORs structural dataset based on the structural characteristics of ORs. The algorithm analyzes the hydrophilicity and hydrophobicity of amino acid sequences, weights the potential transmembrane amino acid residues, compares them with the transmembrane domains of template protein, and then performs local sequence comparison. After experimental validation, the algorithm improved the sequence comparison accuracy of the OR. 2. Homology modeling for ORs and molecular docking experiments between ORs and perfume allergens Two experimentally identified potential perfume allergens, lyral and coumarin, are each recognized by OR10J5 and OR5P3. We performed homology modeling for OR10J5 and OR5P3 separately using a modified homology modeling approach, and then performed molecular docking simulations of both allergens and their ligands. As a consequence, we identified some key potential binding sites, which provided guiding support for the subsequent toxicological and pharmaceutical experiments. 3. Traditional machine learning methods show high dependence on features when solving classification problems For the ORs classification, the accuracy of existing studies is about 90%. In this research, a neural network-based classification model for ORs was proposed. This model combines convolutional neural network, Embedding, MaxPooling, and long and short-term memory, and finally achieves a classification accuracy of over 97%. 4. The emergence of AlphaFold2 has a profound impact on predicting the interaction between protein and ligand docking. The advent of AlphaFold2 brought a lot of high-precision predicted structures of ORs. We performed ligand docking experiments on hydroxycitronellal and Olfr16 provided by AlphaFold, and after comparing the results of the ligand docking experiments between hydroxycitronellal and Olfr16 obtained by traditional homology modeling methods, we found common predicted docking sites for both, LEU37 and LEU144. Investigating the similarities and differences in docking results between these two experimental approaches will help us to further discover how AlphaFold2 will advance ORs research.
- Subject
- olfactory receptors; molecular docking; homology modeling; thesis by publication
- Identifier
- http://hdl.handle.net/1959.13/1481819
- Identifier
- uon:50811
- Rights
- Copyright 2022 Pu Wang
- Language
- eng
- Full Text
- Hits: 394
- Visitors: 614
- Downloads: 263
| Thumbnail | File | Description | Size | Format | |||
|---|---|---|---|---|---|---|---|
| View Details Download | ATTACHMENT01 | Thesis | 6 MB | Adobe Acrobat PDF | View Details Download | ||
| View Details Download | ATTACHMENT02 | Abstract | 417 KB | Adobe Acrobat PDF | View Details Download |