Statistical Test-Based Evolutionary Segmentation of Yeast Genome

Segmentation algorithms emerge observing fluctuations of DNA sequences in alternative homogeneous domains, which are named segments [1]. The key idea is that two genes that are controlled by a single regulatory system should have similar expression patterns in any data set. In this work, we present a new approach based on Evolutionary Algorithms (EAs) that differentiate segments of genes, which are represented by its level of meiotic recombination1. We have tested the algorithm with the yeast genome [2][3] because this organism is very interesting for the research community, as it preserves many biological properties from more complex organisms and it is simple enough to run experiments. We have a file with about 6100 genes, divided into sixteen yeast chromosomes (N). Each gene is a row of the file. Each column of file represents a genomic characteristic under specific conditions (in this case, only the activity of meiotic recombination). The goal is to group consecutive genes properly differentiated from adjacent segments. Each group will be a segment of genes, as it will maintain the physical location within the genome. To measure the relevance of segments the Mann–Whitney statistical test has been used. Each individual of the population is an array of natural numbers with size C, and it represents a collection of cutpoints within the yeast genome. Fifteen of these cutpoints correspond to the boundaries of the sixteen chromosomes of the yeast genome, and they are permanent. The sixteen cutpoints corresponding to centromeres also are permanent, so we have 31 constant cutpoints. The centromere is approximately in the middle of a chromosome and separates it in two branches (L and R). Although these fixed cutpoints (FC=31) cannot be moved, they have been included in all of the individuals, making easier the computational process. For example, if a cutpoint array includes the values 34, 57, 7, 25 and 80, it means that there is a cutpoint between the 34 and the 35 genes, between the 57 and the 58 genes, between the 7 and the 8 genes, etc. We have chosen the Mann-Whitney test as the fitness function. The Mann–Whitney test, also known as the Wilcoxon rank sum test, is a non–parametric test used to test for difference between the medians of two independent groups. This test is the non–parametric equivalent of the two–sample t–test. No distributional assumptions are required for this test, so the test does not assume that the populations follow Gaussian distributions. The choice of this method is due to the necessity of differentiating adjacent segments clearly. If we choose the mean as