Previous Projects

Post-Doc

Roslin Institute, Roslin, Scotland.

The project is financed by a post-doctoral fellowship from Knut and Alice Wallenbergs foundation. I am involved in several subprojects during my time at Roslin Institute.

Mapping of epistatic QTL in Experimental Populations:

We reanalyzed data from several QTL mapping projects managed by Paul Hocking and Dave Burt at Roslin Institute, Gudrun Brockmann at FBN in Germany and Robert W Williams at University of Tennessee. The work was done to evaluate the importance of epistasis in regulating a wide range of mult-factorial traits. For these analyses, we used an SGI Origin 3000 supercomputer at CSAR.

Efficient numerical methods for multi-dimensional genome-scans:

The project implemented and evaluated new numerical methods for mapping of multiple QTL. The methods were shown to improve the power of QTL mapping by, in a numerically efficient way, perform multidimensional searches for QTL in the genome. The project was conducted in collaboration with Kajsa Ljungberg and Sverker Holmgren at the Department of Information Technology at Uppsala University, Sweden.

Methodology for dissecting the genetic regulation of gene-expression:

In this project, we explored the properties of statistical methods used for dissection of the genetic regulation of gene-expression (“Genetical Genomics”). We used experimental data from the first publicly available dataset of this kind as well as simulated data to evaluate existing methods and propose new analytical strategies for this.

 

PhD Student

Dept of Animal Breeding and Genetics, SLU, Uppsala, Sweden.

My thesis presents and discusses the use of various genetic models, high performance computing, global optimization algorithms and statistical methods for mapping Quantitative Trait Loci (QTL). The aim of the work was to develop statistically powerful and computationally efficient methods to detect genomic loci affecting multifactorial traits, and use the methods use to analyse experimental data.

Imprinting is an epigenetic phenomena which causes differential expression of alleles based on their parental origin. A genetic model handling imprinting was used during QTL mapping in an experimental Wild Boar x Large White intercross. The analyses revealed a paternally imprinted QTL with large effect on the development of muscle mass [1].

Parallel computing algorithms for interval mapping and randomization testing in QTL mapping are described. New randomization testing schemes are now computationally feasible due to these algorithms. Selection of appropriate kernel algorithms for solving least squares type problems in QTL mapping is discussed. The importance of optimization of QTL mapping software is also illustrated [2].

A genetic algorithm was shown to be efficient in a multidimensional search for interacting QTL. The genetic algorithm significantly decreases the computational demand when employing simultaneous mapping of multiple QTL, and makes randomization testing based on multidimensional searches computationally feasible [3]. A new randomization testing scheme based on simultaneous mapping of epistatic QTL was also proposed and evaluated. A simulation study showed that the method increases the power to detect epistatic QTL [4].

A large intercross was derived between Red junglefowl and White Leghorn chickens. A number of QTL affecting growth was revealed using the newly developed method for simultaneous mapping of epistatic QTL pairs. In total, 21 QTL were identified, and eleven of these were only detected by the new simultaneous mapping method. Epistasis was shown to be an important component in the genetic regulation of the growth process [5], [6].

Latest update: 03/07/02 14:45

back