EThesis 72 views 24 downloads
Development of a comprehensive 3D structural database of human pan-proteomic interaction with the most valuable sold drugs / William H. Edwards
PDF | Redacted version - open accessDownload (12.66MB)
DOI (Published version): 10.23889/Suthesis.52409
Drug discovery is undertaken to discover new candidate medicines. Identifying new therapeutics is of critical medical, social and economic importance. In recent years the rising cost associated with drug discovery and development has necessitated the more frequent use of in silico approaches. The pr...
No Tags, Be the first to tag this record!
Drug discovery is undertaken to discover new candidate medicines. Identifying new therapeutics is of critical medical, social and economic importance. In recent years the rising cost associated with drug discovery and development has necessitated the more frequent use of in silico approaches. The prediction of protein-ligand interactions using in silico approaches has become widely used to study biomolecular interactions and mechanisms. These approaches allow for high throughput virtual screening of thousands of potential drug candidates at specific protein drug targets. While the data generated provides ample opportunity for scientific and clinical exploitation, challenges are presented concerning the vast quantity of information generated and the ability to utilise this information fruitfully. In this study in silico approaches are used to “virtually screen” the entire human proteome against the 20 most valuable sold drugs in the UK. Homology and protein threading approaches were used for protein structure modelling; AutoDock Vina and DOCK 6.0 were used for protein-ligand docking with the use of High-Performance Computing (HPC). The large-scale application of these approaches was evaluated, and methods iteratively refined to improve predictive accuracy. A novel combinational forecasting method was developed to increase the accuracy of the predictions of the docking programs. The method produced a docking an overall accuracy of 77.05% for identifying known protein interactions and known protein misses correctly. A platform system was developed to allow the vast amount of data to be efficiently reported and visualised within a Graphical User Interface (GUI). The developed database and system prototype have the potential to change the way drugs are developed in the drug discovery sector. This system is a powerful tool which can be used for the advancement of personalised medicine with incorporation of further knowledge of protein interactions as well as the effects of protein variation. The work carried out in this project has contributed towards the development of a comprehensive in silico platform, Human3DProteome (human3dproteome.com), that utilises the system architecture, methodologies, data and methods of data analysis advanced in this project. Human3DProteome is the first public platform which aims to catalogue structural models for every protein in the human body alongside a comprehensive database of predicted small molecule interactions of interest.
Drug Screening, 30 Molecular Structure, Genome, Database
Swansea University Medical School