Martino, Sam Alexander;
(2025)
Applications of Machine Learning and General-Purpose GPU programming in Computational Drug Discovery.
Doctoral thesis (Ph.D), UCL (University College London).
Preview |
Text
SamMartinoThesisCorrected.pdf - Accepted Version Download (29MB) | Preview |
Abstract
Recent advances in parallel computing and machine learning are revolutionising the computational techniques used to investigate potential protein drug targets and identify compounds capable of modulating them. This work provides an overview of the context surrounding this revolution, along with an in-depth discussion of my work applying and developing these methods. It introduces the theory behind graphics processing units, machine learning, and molecular dynamics. Further work is categorised into 2 chapters, focusing on applications of these for investigating targets or screening ligands. They contain 5 Case Studies relating to various stages of drug discovery, highlighting my research contributions to the area. Target Identification and Investigation introduces methodologies used to analyse both experimental and simulated protein data, before focusing on 3 projects in this area. Case Study 1 focuses on applying variational autoencoders to learn the conformational space of BRAF activation segments from experimental structures. Case Study 2 analyses kinetic data on NSP13, a protein vital to SARS-CoV-2, and attempts to simulate the systems critical translocation function. Case Study 3 introduces a novel graph neural network based strategy able to optimise a Markov chain clustering by preserving Kemeny's constant. Large scale screening covers working with small molecules at large scales, explaining how they are processed and evaluated for potential as drugs. Case Study 4 combines several of the introduced methods to build a drug discovery pipeline aimed at finding compounds for NSP13 in the CACHE2 drug discovery challenge, resulting in experimental confirmation of a novel binder. Case Study 5 takes this pipeline and applies it to RAS, another prolific oncogene, and evaluates the beneficial performance of active learning strategies in improving ligand selection.
| Type: | Thesis (Doctoral) |
|---|---|
| Qualification: | Ph.D |
| Title: | Applications of Machine Learning and General-Purpose GPU programming in Computational Drug Discovery |
| Open access status: | An open access version is available from UCL Discovery |
| Language: | English |
| Additional information: | Copyright © The Author 2025. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request. |
| UCL classification: | UCL UCL > Provost and Vice Provost Offices > UCL BEAMS UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Maths and Physical Sciences UCL > Provost and Vice Provost Offices > UCL BEAMS > Faculty of Maths and Physical Sciences > Dept of Physics and Astronomy |
| URI: | https://discovery.ucl.ac.uk/id/eprint/10210514 |
Archive Staff Only
 |
View Item |
