Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.495580
Title: Data exploration and knowledge extraction : their application to the study of endocrine disrupting chemicals
Author: Roncaglioni, Alessandro
ISNI:       0000 0004 2669 9411
Awarding Body: Open University
Current Institution: Open University
Date of Award: 2008
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
Interest in computer-aided methods for investigating the biological field has increased significantly. One method is Quantitative Structure-Activity Relationships (QSAR), a valuable technique for predicting the effects of a substance from its chemical structure. A challenging application of QSAR is in characterizing the (bio)activity profiles of chemicals. Endocrine disrupters (EDs) are exogenous substances interfering with the function of the endocrine system and represent an interesting field of application for in silico methods. EDs targets include nuclear receptors, particularly effects mediated by the oestrogen receptor (ER). They are also mentioned as substances requiring a more detailed control and specific authorisation within REACH, the new European legislation on chemicals. QSAR represents a challenging method to approach data gap about EDs since REACH substantially boosted interest on computational chemistry to replace experimental testing. This work: aimed to explore the status, availability and reliability of non-testing methods applied to endocrine disruption via oestrogen receptors and eventually to propose new models easily exploitable in regulatory contexts. The work evaluated existing QSAR models present in literature to assess their validity on the basis of the OECD principles for QSAR validation. Different kinds of models have been analysed and they were externally validated with new data found in the literature. Furthermore, new QSAR binary classifiers have been developed using different data mining techniques (e.g.: classification trees, fuzzy logic, neural networks) based on a very large and heterogeneous dataset of chemical compounds. The focus was given to both binding (RBA) and transcriptional activity (RA) better to characterize the effects of EDs. A possible combination of the models was also explored. A very good accuracy was achieved for both RA and RBA (>85%). These models can be a valuable complement to in vivo and in vitro studies in the toxicological characterisation of chemical compounds.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.495580  DOI:
Share: