| dc.contributor.advisor | Bickeböller, Heike Prof. Dr. | |
| dc.contributor.author | Wendel, Bernadette | |
| dc.date.accessioned | 2023-02-23T14:58:38Z | |
| dc.date.available | 2023-03-30T00:50:12Z | |
| dc.date.issued | 2023-02-23 | |
| dc.identifier.uri | http://resolver.sub.uni-goettingen.de/purl?ediss-11858/14531 | |
| dc.identifier.uri | http://dx.doi.org/10.53846/goediss-9750 | |
| dc.format.extent | 117 Seiten | de |
| dc.language.iso | eng | de |
| dc.rights.uri | http://creativecommons.org/licenses/by/4.0/ | |
| dc.subject.ddc | 610 | de |
| dc.title | Genome-wide association studies and follow-up kernel approaches in the longitudinal PsyCourse Study | de |
| dc.type | doctoralThesis | de |
| dc.contributor.referee | Beißbarth, Tim Prof. Dr. | |
| dc.date.examination | 2023-02-17 | de |
| dc.description.abstracteng | A genetic association study is a popular method to analyse the connection of genomic factors with disorders or disease-related phenotypes. There are various study types including genome-wide association studies (GWASs) and pathway analyses. In the simplest case, the studied phenotype is either a case-control status or a quantitative trait, e.g., a cognitive test score, with one measurement per individual. This thesis focuses on GWASs and pathway analyses for longitudinal phenotypes in which multiple correlated phenotype measurements per individual are available. The focal phenotypes of this thesis are a group of essential cognitive functions in the longitudinal PsyCourse Study, the executive functions (EFs).
Longitudinal GWAS requires special statistical methods, in which hundreds of thousands of single nucleotide polymorphisms (SNPs) are tested for association with a longitudinal phenotype. Linear mixed models (LMMs) are one popular option to model the correlation structure of the multiple assessments with random effects. LMMs can also handle missing measurements, a frequently occurring problem with longitudinal data. Moreover, LMMs are connected with kernel machine regression (KMR) analyses, which are based on kernel methods. We can apply KMR to perform a pathway analysis, in which a whole pathway (or gene set) is tested for association. These kernel methods can handle high-dimensional genetic data by transforming the data into a lower-dimensional similarity matrix. This similarity matrix or kernel matrix describes the genetic similarities of every pair of study subjects and can be modelled very flexibly. Thus, we are able to integrate additional biological aspects into the kernel, e.g. network information.
This thesis begins with conducting a longitudinal GWAS, in which we aim to identify SNPs influencing the short-term course of EFs. We apply LMMs to study the course over time of EFs. We use data from the PsyCourse Study, in which EFs are assessed at multiple measurement points with cognitive tests, e.g., the Trail Making Test, part B (TMT-B). Nine highly correlated genome-wide significant SNPs are identified as being associated with the change over time in TMT-B. This result is replicated in an independent sample.
The main objective of this thesis is the extension of KMR to long-KMR to enable the performance of a longitudinal pathway analysis. We include additional random effects to KMR to create long-KMR. Long-KMR is further able to integrate network information by utilising a network-based kernel and thus can be applied as a topology-based pathway analysis. Moreover, long-KMR is able to model a pathway as a main genetic and/or genetic-time-interaction effect, either of which can be tested for association. The genetic-time-interaction effect allows studying the association of a pathway with the time course of a phenotype. Overall, long-KMR demonstrates a higher power compared to another longitudinal KMR method previously developed. The power increases further when applying the network kernel to include biological information. Long-KMR is available as an R package kalpra. | de |
| dc.contributor.coReferee | Kneib, Thomas Prof. Dr. | |
| dc.subject.eng | Longitudinal genome-wide association study | de |
| dc.subject.eng | Longitudinal pathway analysis | de |
| dc.subject.eng | Kernel machine regression | de |
| dc.subject.eng | Network kernel | de |
| dc.subject.eng | PsyCourse Study | de |
| dc.subject.eng | Executive Functions | de |
| dc.identifier.urn | urn:nbn:de:gbv:7-ediss-14531-6 | |
| dc.affiliation.institute | Medizinische Fakultät | de |
| dc.subject.gokfull | Medizin (PPN619874732) | de |
| dc.subject.gokfull | Medizinische Statistik / Biometrie / Epidemiologie - Allgemein- und Gesamtdarstellungen (PPN619875046) | de |
| dc.subject.gokfull | Humangenetik (PPN619875267) | de |
| dc.description.embargoed | 2023-03-30 | de |
| dc.identifier.ppn | 1837640114 | |
| dc.notes.confirmationsent | Confirmation sent 2023-02-23T15:15:01 | de |