UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

Use of neural networks to model molecular structure and function

Hirst, Jonathan Darrell; (1993) Use of neural networks to model molecular structure and function. Doctoral thesis (Ph.D), UCL (University College London). Green open access

[thumbnail of Use_of_neural_networks_to_mode.pdf]

Download (8MB) | Preview


This thesis is a study of some applications of neural networks - a recent computer algorithm - to modelling the structure and function of biologically important molecules. In Chapter 1, an introduction to neural networks is given. An overview of quantitative structure activity relationships (QSARs) is presented. The applications of neural networks to QSAR and to the prediction of structural and functional features of protein and nucleic acid sequences are reviewed. The neural network algorithms used are discussed in Chapter 2. In Chapter 3, a two-layer feed-forward neural network has been trained to recognise an ATP/GTP-binding local sequence motif. A comparably sophisticated statistical method was developed, which performed marginally better than the neural network. In a second study, described in Chapters 4 and 5, one of the largest data sets available for developing a quantitative structure activity relationship - the inhibition of dihydrofolate reductase by 2,4-diamino-6,6-dimethyl-5-phenyldihydrotriazine derivatives has been used to benchmark several computational methods. A hidden-layer neural network, a decision tree and inductive logic programming have been compared with the more established methods of linear regression and nearest neighbour. The data were represented in two ways: by the traditional Hansch parameters and by a new set of descriptors designed to allow the formulation of rules relating the activity of the inhibitors to their chemical structure. The performance of neural networks has been assessed rigourously in two distinct areas of biomolecular modelling; sequence analysis and drug design. The conclusions of these studies are presented in Chapter 6.

Type: Thesis (Doctoral)
Qualification: Ph.D
Title: Use of neural networks to model molecular structure and function
Open access status: An open access version is available from UCL Discovery
Language: English
Additional information: Thesis digitised by ProQuest.
Keywords: Biological sciences; Model; Molecular structure and function; Neural networks
URI: https://discovery.ucl.ac.uk/id/eprint/10103750
Downloads since deposit
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item