Statistical Genetics

Statistical Genetics

The following research topics within statistical genetics and bioinformatics are of particular interest at the department.

Forensic genetics and DNA-investigations are an increasingly important part of e.g. immigration-, paternity- and criminal cases. There is an international consensus about investigation of Short Tandem Repeat (STR) regions with polymerase chain reaction based methods, standardized by international collaboration. The question of quantification and presentation of the weight of the DNA evidence for the court contains many statistical and philosophical statistical problems, as does the investigation of the behaviour of the standardized kit’s used in the DNA- analysis. Research in forensic statistics has to acknowledge the special world of decision-making in the courtroom, and it takes methods and principles from probability theory, classical likelihood theory and Bayesian networks. Mikkel Meyer Andersen, Ellen Susanne Christensen, Poul Svante Eriksen, Torben Tvedebrink and Søren Vilsen have contributed on various aspects in research on forensic statistics.

Quantitative genetics: Traits (phenotypes) for a group of animals is traditionally modeled using a mixed model where the mean depends on genetic random effects whose correlation structure is determined by a pedigree for the animals. Using this model one can for example in breeding programmes select the animals whose predicted genetic effects are favorable for the target of the breeding programme. Using techniques of computational statistics, Rasmus Waagepetersen has worked on an extended model where also the residual variance is under genetic control. This is of interest if a selection target is reduced variance of a trait. Recent advances in quantitative genetics build on data on the molecular level where individuals are genotyped at markers located on the DNA and interest then focuses on building predictive models based on the marker genotypes.