Title Page
Contents
List of Abbrevations 7
BACKGROUND 8
1. Descriptor 8
2. Topological water network concept for new protein-based approaches 10
3. Integer type fingerprint concept for new ligand-based approaches 12
PART I. New protein-based approaches using topological water network 13
I.1. Introduction 13
I.2. Materials and Methods 17
I.2.1. Preparation and capping of all amino acids 17
I.2.2. Molecular Dynamics (MD) Simulation 20
I.2.3. Topological Water Network (TWN) Analysis 21
I.2.4. Analysis of PDB structures 23
I.3. Results and Discussion 25
I.3.1. Validation of MD simulation results for each amino acid 25
I.3.2. TWN Pattern around amino acids 29
I.3.3. TWN Pattern for Residues of the PDB Structures 46
I.4. Conclusion 56
PART II. New ligand-based approaches using chemical fingerprint 57
II.1. Introduction 57
II.2. Materials and Methods 60
II.2.1. Dataset 60
II.2.2. Fingerprint Calculation 62
II.2.3. Model Building 63
II.2.4. Model Evaluation 64
II.3. Results 65
II.3.1. Dataset Splitting 65
II.3.2. Model Building and Evaluation 68
II.3.3. External Validation 74
II.4. Discussion 81
II.5. Conclusion 84
CONCLUSION 93
REFERENCES 95
ABSTRACT IN KOREAN 104
Table 1. The number of 3, 4, 5, and 6-ring TWNs observed around various atoms of all... 33
Table 2. The number of TWNs observed around various atoms of all amino acids in the MD... 40
Table 3. The number of TWNs observed around various atoms of all amino acids in all the... 48
Table 4. Dataset information used in the hERG blocker prediction study. 61
Table 5. Models built using integer type fingerprints. 70
Table 6. Models built using binary type fingerprints. 72
Table 7. Model prediction for external set-1 (Ex-1) and external set-2 (Ex-2) using... 77
Table 8. Model prediction for external set-1 (Ex-1) and external set-2 (Ex-2) using... 79
Table 9. Model prediction including further parameters for external set-1... 85
Table 10. Model prediction including further parameters for external set-1... 87
Table 11. Model prediction including further parameters for external set-2... 89
Table 12. Model prediction including further parameters for external set-2... 91
Figure 1. Various descriptors used as protein- and ligand- based computational researches. 9
Figure 2. Topological water network (a) and their application for drug design (b) 11
Figure 3. Fingerprint types representing ligand structure 12
Figure 4. 2D Structures of the amino acids studied in the present work. 18
Figure 5. Amino acid capping at both termini. 19
Figure 6. Hydropathy index of all amino acids according to the Kyte-Doolittle scale. 25
Figure 7. Ramachandran plots for the representative amino acids depending on hydrophobicity. 27
Figure 8. RMSD plot for the backbone atoms of amino acids from the initial structures as... 28
Figure 9. An illustration of a 3-ring TWN closely located near the O atom of an amino acid... 32
Figure 10. An illustration of a 3-ring TWN closely located near one of the N atoms of... 47
Figure 11. Principle component analysis (PCA) for the training and test sets of hERG blockers. 66
Figure 12. Fingerprint frequency calculation between the training and test sets of hERG blockers. 67
Figure 13. Models built using integer and binary type fingerprints for hERG blockers. 73
Figure 14. Model prediction to external sets. 80
Figure 15. (a) protein based- and (b) ligand-based new approaches in this research 94