Use this URL to cite or link to this record in EThOS:
Title: Statistical models for the genetic analysis of longitudinal data
Author: Jaffrezic, Florence
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 2001
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
The first objective of this work was to compare and contrast different methodologies for genetic analysis. As the range of all possible models can be very large in practice, it is advisable to have a preliminary idea of the covariance structure of the data, and a non-parametric approach based on the variogram was proposed. It is especially adapted for exploratory analysis when a large number of observations is available per subject over time and was applied to the analysis of daily records for milk production in dairy cattle. Model comparisons in the univariate case showed that character processes were generally better able to fit the covariance structure than random regression with fewer parameters. However, CP models do not allow a straightforward extension to the multivariate case. Further research showed that structured antedependence models offer similar advantages to character processes compared to random regression while allowing an extension to multi-trait analyses. SAD models were even able to capture the highly non-stationary correlation pattern in the application to lactation curve analysis. For genetic evaluation of dairy cattle, longitudinal models can easily provide estimation of individual cumulative milk productions as well as genetic values at 305 days. However, these predictions do not take into account the drying-off process and can be highly overestimated for short lactations. A methodology to correct them was suggested. All these analyses were performed in the case of normally distributed longitudinal data. An extension to the genetic analysis of non-normally repeated measures was considered. Estimation procedure becomes much more complicated and requires the use of Markov Chain Monte Carlo methods. In this study antedependence models appeared to be the most appropriate for genetic analysis of longitudinal data.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available