Use this URL to cite or link to this record in EThOS:
Title: Statistical methods to infer kinase activities and kinase-substrate interactions using phosphoproteomic data
Author: Hernandez Armenta, Claudia Ivonne
ISNI:       0000 0004 9348 0962
Awarding Body: University of Cambridge
Current Institution: University of Cambridge
Date of Award: 2020
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
The activity of kinases play an important role during cell signalling. They phosphorylate proteins to regulate their interactions, structural conformation, cell location and enzymatic activation. Mass Spectrometry (MS) based phosphorylation studies (i.e. phosphoproteomics) has revolutionized the study of kinase signalling, allowing for the quantification of thousands of phosphorylated protein sites. However only a small fraction of these events are associated with a regulatory kinase. In this work, different methods to predict kinase activity and to inter new substrates are tested with the objective of contributing to fill this gap in knowledge. I first present the comparison and evaluation of algorithms to predict kinase activity profiles using a compilation of human published phosphoproteomics datasets where the regulation of the kinases is well characterized. For this I tested the impact of the following variables in the best performing approaches: the integration of kinase binding specificities, the number of known kinase targets, and the type of experimental evidence supporting each kinase-substrate relationship. I then worked with a larger set of phosphoproteomic experiments to infer phospho-coregulation patterns between kinase activation and protein phosphorylation levels. The main objective was to test the predictive performance of new kinase-substrate logistic predictors that use phospho-coregulation information with kinase binding specificity and protein-protein functional associations. The most predictive feature was the kinase binding specificity for the majority of the kinases tested, however the combined predictors using protein-protein associations and the phospho-coregulation scores showed a modest improvement for a set of kinases. Finally, I analyzed time course phosphoproteomic experiments to identify the regulatory mechanisms occurring during different stages of yeast meiosis. For this purpose, I inferred at different time points the changes in kinase activity and phosphorylation levels inside protein complexes. These findings were integrated with a second temporal screening of yeast strains harboring an inactive Cdc5 kinase to identify their regulated targets. The Cdc5 candidate substrates were inferred using the phospho-coregulation information across a yeast atlas conformed of 217 biological conditions, and further evaluation was made analysing the overlap between the functional annotation of the candidates and the prior knowledge on Cdc5 genetic experiments. The methods developed in this work can be combined with quantitative and genetic experiments to study kinase signalling and determine the mechanisms governing cell behaviour and cell fate in different diseases and specific cellular processes. This knowledge is crucial to identify potential biomarkers for clinical diagnosis or candidate targets for drug development.
Supervisor: Beltrao, Pedro Sponsor: EMBL
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
Keywords: kinase signalling ; phospho-proteomics ; computational biology ; kinase-targets ; predictive methods ; mass-spectrometry