Protein design using artificial intelligence

The emergence of new and resistant strains of pathogens, coupled with the slow and expensive process of drug discovery, has created an urgent need for innovative approaches in pharmaceutical development. This thesis explores how machine learning can be leveraged to accelerate the discovery and design of new pharmaceuticals. Each chapter addresses different aspects of the drug design process, proposing software solutions aimed at enhancing the productivity of researchers in the field. The contributions section of the thesis begins with the examination of the fundamental problem of protein folding, focusing on AlphaFold and its unprecedented accuracy in predicting protein structures, which rivals experimental methods. The chapter discusses both the innovations brought forth by these models and their technical limitations. Following this, the thesis pivots to the issue of Molecular Recognition, where protein conformation plays a critical role. This challenge is explored through the Puzzle Solver project, which uses a jigsaw puzzle analogy to approach the molecular recognition problem. The exploration of protein predictors led to identifying a key issue in pharmaceutical research: the need for reverse protein prediction, specifically parametric protein design. This is addressed through the SilverSurfer and PS-GO projects. The final section of the thesis focuses on integrating cutting-edge tools and simplifying workflows for pharmaceutical researchers. The PROFASA platform offers a single point for protein design needs, while ChemFlow lowers the technical knowledge barrier to designing workflows. This allows researchers to more easily utilise state-of-the-art tools and focus on their area of expertise.

Grantor dc:publisher

University College Cork

Year dc:date.issued

2025

Author dc:creator

• Marcu, Stefan-Bogdan

Advisors dc:contributor.advisor

• Tabirca, Marius-Sabin
• Tangney, Mark

• AI
• Machine learning
• Protein design
• Transformer
• Protein folding

Statement dc:rights

• © 2025, Stefan-Bogdan Marcu.

Licence dc:rights.uri

• https://creativecommons.org/licenses/by/4.0/

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/10468/17423

OAI identifier oai:identifier

oai:cora.ucc.ie:10468/17423