Unwarping and Analysing Electrophoresis Gels

Doctoral thesis, 2005

Gel electrophoresis is a technology in molecular biology that is used to separate and quantify biomolecules, for example DNA and proteins. Ideally the position in a gel image corresponds to one or several properties of the molecules, such as size and net charge, but in practise the separation pattern is often warped, that is, geometrically distorted. This thesis presents methods for unwarping and analysing electrophoresis gel images. There are three main contributions in the thesis: (i) An image analysis method for simultaneous unwarping and classification of multi-track 1-D gels using a database of track profiles. The main advantage of this method is that the two analysis steps are done automatically and simultaneously. The method is motivated by and applied to pulse-field electrophoresis gels from a genotyping of potentially antibiotic-resistant bacteria. (ii) A physical model and methods for individual unwarping and pair-wise alignment of 2-D gel images. Here, the main advantages are an explicit control of image deformations and that the pair-wise alignment is automatic. The methods are evaluated on a set of protein gels from a study of the adaptation of yeast cells to saline conditions. The image alignment method is also applied and found to be robust in a comparative study of different yeast strains where the protein spot pattern differs substantially between the strains. (iii) The final main contribution is a statistical exploration of variation in quantitative 2-D gel data from three gel sets. Here, parallels to the design and analysis of microarray experiments are discussed and utilised.