Please use this identifier to cite or link to this item:
Full metadata record
DC FieldValueLanguage
dc.contributor.authorAlhossary, Amr Ali Mokhtaren
dc.identifier.citationAlhossary, A. A. M. (2018). Accurately accelerating drug design workflow. Doctoral thesis, Nanyang Technological University, Singapore.en
dc.description.abstractIn a world where pathogenic bacteria, viruses, as well as cancer develop resistance to drugs on faster pace than discovering new ones, researchers bear the heaviest weight to design new drugs to overcome such phenomenon of drug resistance. Drug design is one of the most challenging tasks in computational and structural biology, which aims at developing new drugs or enhancing currently known drugs against certain diseases based on the knowledge of a biological target. This thesis is about accelerating drug design work flow, through accurate acceleration of molecular docking tools, proper selection of candidates for Multiple Receptors Conformation (MRC) docking, and molecular dynamics (MD) simulation. In this work, I first developed QuickVina 2, a fast, accurate, and reliable molecular docking tool that depends on the powerful scoring function of AutoDock Vina and accelerated search of QuickVina. QuickVina 2 was tested against the 195 protein-ligand complexes of the core set of PDBbind 2014, using default exhaustiveness level of 8. It successfully attained up to 20.49-fold acceleration over Vina with tendency for higher acceleration when the number of dimensions/variables increases. Meanwhile, 70% of its predicted modes were equal to or better than original Vina in terms of binding energy. The remaining 30% had average Energy difference only 0.58 Kcal/mol. The Pearson’s correlation coefficient (r) between AutoDock Vina’s and QuickVina 2’s binding energy was 0.967 for the first predicted mode and 0.911 for the sum of all predicted modes. QuickVina 2 was found to be more accurate than GOLD 5.2 and is only slightly less accurate than Dock 6.6. QuickVina 2 was employed to propose drug fragments for Dengue Virus Non-Structure Protein 5 (DENV-NS5), and the result was compared to AutoDock Vina result as a measure of double confirmation. Both QuickVina 2 and AutoDock Vina detected the same 13 fragments with slight differences in their estimated binding energies while QuickVina 2 detected three additional fragments. Two of the fragments were subjected to MD simulations for in silico validation. The simulation results suggest that the proposed ligands are plausible and could be considered for further computational and experimental validation, as well as lead optimization. The work also involved refining the selection criteria of receptor conformation candidates that undergo MRC docking, in order to ensure diversity and increase sensitivity (decrease false negative rate) of detection. QuickVina 2 was taken then to another dimension by enabling it to search wide search spaces, after introducing inter-process spatio-temporal integration between the searching threads to communicate their collective wisdom. That work resulted in the release of QuickVina-W, a tool suitable for Blind Docking. QuickVina-W explores four folds the number of points that Vina explores, in a more efficient way. It proved to be faster than QuickVina 2 (with average and maximum normalized overall time accelerations of 3.60 and 34.33 folds in relation to Vina versus 1.98 and 18.02 respectively), yet better than AutoDock Vina in terms of binding energy (78% of predictions with binding energy better than or equal to Vina) and RMSD (Root Mean Square Distance) to experimental data (with success rate of 72% by QuickVina-W versus 63% by Vina). It was based on the observation that the Average Sum of Proximity relative Frequencies (ASoF) of searching threads is ever increasing with search progression, and on the theory that allowing a searching thread to communicate with other nearby threads to make use of their wisdom, would increase the speed and sensitivity of that searching thread, in a way relevant to the increasing ASoF. This work monitored the ASoF and proved its direct relation to decision taking increased speed and accuracy which are reflected in turn on the search process.en
dc.format.extent157 p.en
dc.subjectDRNTU::Engineering::Computer science and engineering::Computer applications::Life and medical sciencesen
dc.subjectDRNTU::Science::Biological sciences::Molecular biologyen
dc.titleAccurately accelerating drug design workflowen
dc.contributor.supervisorKwoh Chee Keongen
dc.contributor.supervisorMu Yuguangen
dc.contributor.schoolSchool of Computer Science and Engineeringen
dc.description.degreeDoctor of Philosophyen
item.fulltextWith Fulltext-
Appears in Collections:SCSE Theses
Files in This Item:
File Description SizeFormat 
Amr ALHOSSARY G1201969D Thesis (Final).pdfMy PhD thesis7.47 MBAdobe PDFThumbnail

Page view(s)

Updated on Aug 17, 2022

Download(s) 10

Updated on Aug 17, 2022

Google ScholarTM




Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.