Statistical inference from large-scale genomic data

摘要: This thesis explores the potential of statistical inference methodologies in their applications functional genomics. In essence, it summarises algorithmic findings this field, providing step-by-step analytical for deciphering biological knowledge from large-scale genomic data, mainly microarray gene expression time series. This covers a range topics investigation complex multivariate data. One focus involves using clustering as method and another is cluster validation to extract meaningful information Information gained application these various techniques can then be used conjointly elucidation regulatory networks, ultimate goal type analysis. First, new tight data proposed obtain tighter potentially more informative clusters. Next, fully utilise validation, validity index defined based on one most important ontologies within Bioinformatics community, Gene Ontology. The bridges gap current literature, sense that takes into account not only variations Ontology categories specificities significance clusters, but also structure Finally, Bayesian probability applied making heterogeneous integrated with previous efforts thesis, aim network inference. system comes stochastic process achieve robustness noise, yet remains efficient enough analysis. Ultimately, solutions presented serve building blocks an intelligent interpreting understanding organisation genome.