Use this URL to cite or link to this record in EThOS:
Title: The application of spectral geometry to 3D molecular shape comparison
Author: Seddon, Matthew
ISNI:       0000 0004 7230 8354
Awarding Body: University of Sheffield
Current Institution: University of Sheffield
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
Molecular shape has long been recognised as a key determinant of molecular interactions (Nicholls et al., 2010). Several methods have been developed to represent the shapes of molecules, however, their performance is inadequate for large-scale virtual screening both in effectiveness and throughput (Maggiora, Vogt, Stumpfe, & Bajorath, 2014). In general, this has been attributed to computational complexity with regards to finding the optimal alignment and bioactive conformation of the small molecules. This work addresses this problem by applying spectral geometry to three-dimensional molecular shape representation. Spectral geometry has been developed in the field of computer vision for information retrieval of flexible three-dimensional shapes. Typical applications include identifying a shape, such as a human form, in a variety of poses. Of particular interest is the ability to produce 3D shape descriptors that are alignment invariant and capture some notion of flexibility. The main contribution of this thesis is the application of spectral geometry to the domain of 3D molecular shape and the derivation of descriptors suitable for large scale virtual screening. The spectral geometry descriptors are compared to existing shape comparison methods to evaluate their performance for virtual screening. The result is an efficient descriptor that outperforms existing descriptor methods and performs as well as a Gaussian alignment-based approach on some measures.
Supervisor: Gillet, Val ; Packer, Martin Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available