BIN301 Genomic and Pedigree-Based Prediction of Genetic Value
Showing course contents for the educational year 2022 - 2023 .
Course responsible: Gunnar Klemetsdal
Teachers: Jørgen Ødegård, Morten Lillemo, Theodorus Hendrikus Elisabeth Meuwissen
ECTS credits: 10
Faculty: Faculty of Biosciences
Teaching language: EN
Limits of class size:
Teaching exam periods:
This course starts in Autumn parallel. This course has teaching/evaluation in Autumn parallel.
Course frequency: Annually
First time: Study year 2020-2021
Prediction of genetic value using mixed model theory - use of genomic data - use of pedigree data, primarily by use of an animal model, also with more than one random effect and, in addition, multivariate models. Motivation of models used for survival, modelling of categorical data (e.g, for diseases or number of lambs born), longitudinal models, in addition to modelling of genotype by environment interaction. Prediction errors will be calculated.
Introduction to estimation of variance components for random variables.
Examples from plant and animal breeding.
Students will have an overview over state-of-the-art predictions of genetic value for animal and plant populations, and their relevance in breeding programs.
After the course students should be able to predict genetic values, and thus phenotypes, based on genomic and family information (SNP- and G-BLUP, and pedigree based BLUP), with a focus to animal models. Model bias is covered and prediction errors will be calculated. The students will construct mixed model equations for animal models with more than one random effect (which in addition to animal and error might contain permanent environment of animal, litter, or non-additive (gene) effects), in multivariate models, and being introduced to modeling of survival, categorical data (e.g, for diseases or number of lambs born), longitudinal models, in addition to modelling of genotype by environment interaction.
The weekly assignments will be programmed in R, but students will learn commercial software (ASReml), to calculate breeding values, and to estimate variance components, in the most simple models.
Lectures and computer exercises each week (2 h each). Problem-based learning is applied in combination with computer-based exercises. Presentation of predicted genetic values of own data set in a written report, which functions as the single, final evaluation of the course.
Teacher assistance linked to data labs and lectures, and personal assistance, when/if needed.
Lecture notes and selected papers. Exercises, notes, programs, and other material will be made available on the course web site.
STAT200 Regression, or similar course in statistical linear models.
General breeding (HFA200, or BIO248) and similar.
Matrix and vector manipulation (linear algebra). Acquaintance with MS Excel and R or other computer programming.
Presentation of own term paper for the other students, and participation at such presentations.
Handed in term paper, with A-F grading, jointly by external sensor and teacher.
250 hours, including individual study, exercises and participation and presentation in discussion groups / lectures.
Reduction of credits:
5 ECTS reduction for students with exam in HFA301.
Type of course:
Discussion groups / lectures: 2 hours per week. Datalab / exercises: 2 hours per week.
The course was revised in 2019, by including genomic relationships in addition to plant breeding. In 2022, one aim to introduce estimation of variance components, in the simplest models.
An external examiner assesses the term paper.
Examination details: Term paper: Letter grades