Computational reconstruction of non-crossover recombination

Abstract

As a diploid organism, humans inherit two copies of each autosomal chromosome, one from the father and one from the mother. The copy of a chromosome transmitted by a parent is reshuffle, via meiotic recombination, of the two parental copies through meiotic recombination. Most apparent are large chromosome parts of alternating chromosomal origin separated by crossover recombination. The other type of meiotic recombination is gene conversion, small segment of one of the parental homologous chromosomes on the background of the other. Gene conversions arise through process called non-crossover recombination (NCO). Recombinations can only be detected indirectly in offspring, at heterozygous markers, sites where the two homologous chromosomes differ. Sequencing errors, genotyping errors, phasing errors and structural variations can all mimic the signature of gene conversions making them difficult to be detected reliably. Not all NCOs create gene conversions, due to their short span and limit number of heterozygous markers. Large NCOs have a better chance to generate gene conversion than short NCO, due to higher likelihood of overlapping heterozygous markers. This makes it difficult to infer information about NCOs from gene conversions. This thesis makes three main contributions. First, three-generation families are used to find and verify gene conversions showing age and sex dependent patterns for these events. Second, a statistical framework modelling the underlying NCO processes generating these gene conversions is developed in order to obtain the underlying length distribution and their quantity. Finally, combining crossovers and non-crossover recombinations to construct a complete human recombination map revealing their contribution to de novo mutagenesis.