Peter Sørensen


Peter Sørensen
Se relationer på Aarhus Universitet


Molecular prediction of disease and production traits in livestock

The overall research goal is to develop a statistical procedure that identifies which set of biological molecules (e.g., gene transcripts, proteins, and metabolites) best predicts phenotypes for disease and production traits in livestock. In developing this procedure the focus is on statistical methods that

1. account for relationships among biological molecules (e.g. Gaussian graphical modeling)

2. use prior information about the relationship among biological molecules.


We are working on a supervised stochastic search variable selection procedure for identifying promising subsets of molecular predictors of phenotypes in individuals. The procedure uses prior biological information and combines observations from large-scale ‘omics’ data. We have implemented this procedure into a fortran program which is currently being tested.


We are also looking into Gaussian graphical modeling, which is a multivariate statistical technique that can be used to infer relationships among molecular variables such as gene transcripts, proteins, and metabolites.

ID: 1712