New approaches in omics data modelling
The breakthrough in the technological field has allowed the extraction of large amounts of the so-called omics data. The analysis and Integration of this type of data by means of advanced statistical and bioinformatics methods will allow the improvement in the management of diseases. The diversity a...
Gespeichert in:
1. Verfasser: | |
---|---|
Format: | Dissertation |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext bestellen |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | The breakthrough in the technological field has allowed the extraction of large
amounts of the so-called omics data. The analysis and Integration of this type of
data by means of advanced statistical and bioinformatics methods will allow the
improvement in the management of diseases. The diversity and complexity of
omics data has encouraged the development of hundreds of new statistical
methods to meet this objective. Therefore, having the appropriate methods to
accommodate different data distributions and modelling complex data structures
becomes essential. This thesis presents advances in three directions in this
regard. First, the study of several methods to assess non-linear associations
which is relevant when assessing the effect of environmental exposures (i.e
exposome) on complex diseases. The study is accompanied by the
development of the R package nlOmicAssoc. Second, the simplex distribution is
proposed to analyse methylome data since this distribution properly fits beta
values that are generated in this type of studies. The extension to generalized
linear models with simplex response is also proposed. Lastly, an R package,
HOmics, has been developed to incorporate a priori biological knowledge into
association studies by using Bayesian hierarchical models. It also implements
methods to model the dependence between omics data, enabling data
integration
L’avenç en el camp tecnològic ens ha permès obtenir grans quantitats de les
anomenades dades òmiques. L’anàlisi i integració d’aquesta mena de dades
mitjançant mètodes estadístics i bioinformàtics avançats ha de permetre la
millora en el maneig de les malalties. La diversitat i complexitat de les dades
òmiques ha incentivat el desenvolupament de centenars de nous mètodes
estadístics per a complir amb aquest objectiu. Per tant, és primordial disposar
de mètodes que acomodin les distribucions adequades i modelin estructures de
dades complexes. Davant d’això, aquesta tesi presenta avenços en tres
direccions. En primer lloc, l’estudi de diferents mètodes per a analitzar
associacions no lineals, molt rellevant en estudis d’associació entre exposicions
mediambientals (i.e. exposoma) i malalties complexes. Aquesta anàlisi va
acompanyada del desenvolupament del paquet de R nlOmicAssoc. En segon
lloc, es proposa utilitzar la distribució simplex per analitzar dades metilòmiques,
donat que aquesta distribució ajusta els valors beta generats en aquesta mena
d’estudis. També es formula l’extensió a models linea |
---|