Show simple item record

Files in this item


Item metadata

dc.contributor.advisorBarker, Daniel
dc.contributor.advisorIggo, Richard
dc.contributor.authorLim, Ji-Hyun
dc.description.abstractThe tumour suppressor p53 protein plays a central role in the DNA damage response/checkpoint pathways leading to DNA repair, cell cycle arrest, apoptosis and senescence. The activation of p53-mediated pathways is primarily facilitated by the binding of tetrameric p53 to two 'half-sites', each consisting of a decameric p53 response element (RE). Functional REs are directly adjacent or separated by a small number of 1-13 'spacer' base pairs (bp). The p53 RE is detected by exact or inexact matches to the palindromic sequence represented by the regular expression [AG][AG][AG]C[AT][TA]G[TC][TC][TC] or a position weight matrix (PWM). The use of matrix-based and regular expression pattern-matching techniques, however, leads to an overwhelming number of false positives. A more specific model, which combines multiple factors known to influence p53-dependent transcription, is required for accurate detection of the binding sites. In this thesis, we present a logistic regression based model which integrates sequence information and epigenetic information to predict human p53 binding sites. Sequence information includes the PWM score and the spacer length between the two half-sites of the observed binding site. To integrate epigenetic information, we analyzed the surrounding region of the binding site for the presence of mono- and trimethylation patterns of histone H3 lysine 4 (H3K4). Our model showed a high level of performance on both a high-resolution data set of functional p53 binding sites from the experimental literature (ChIP data) and the whole human genome. Comparing our model with a simpler sequence-only model, we demonstrated that the prediction accuracy of the sequence-only model could be improved by incorporating epigenetic information, such as the two histone modification marks H3K4me1 and H3K4me3.en_US
dc.publisherUniversity of St Andrews
dc.subjectRegulatory regionsen_US
dc.subjectLogistic regressionen_US
dc.subject.lcshp53 proteinen_US
dc.subject.lcshBinding sites (Biochemistry)en_US
dc.subject.lcshGene regulatory networksen_US
dc.subject.lcshLogistic regression analysisen_US
dc.titleA computational approach to discovering p53 binding sites in the human genomeen_US
dc.contributor.sponsorBiotechnology and Biological Sciences Research Council (BBSRC)en_US
dc.type.qualificationnamePhD Doctor of Philosophyen_US
dc.publisher.institutionThe University of St Andrewsen_US
dc.publisher.departmentSchool of Medicineen_US

This item appears in the following Collection(s)

Show simple item record