Artificial Neural Networks in the Perception of Genetic Differentiation Caused by Migration

Abstract

Understanding genetic diversity is exceptionally important in order to ensure variability and viability among populations of the same species. Genetic diversity among populations is a consequence of different evolutionary mechanisms that act on them, resulting in different genotypic and allelic frequencies between generations and their future generations; among the different evolutionary mechanisms is migration or gene flow. Self-Organizing Maps (SOM) is an interesting tool to organize and map populations, in addition to highlighting the effects of genetic diversity caused by different evolutionary mechanisms, including migration. The objective of this work was to verify the effects of migration along the generations, analyzing them according to the Conventional Techniques of Biostatistics - Nei, Hedrick and Tocher Cluster Statistics - and, later, to analyze if the self-organizing maps are able to map the effects caused by it. This way, base populations were generated in Hardy-Weinberg equilibrium with 1000 individuals each, 100 codominant diallelic loci, and allelic frequencies equal to p = q = 0.50, which were used to simulate the effects of migration. The simulation is justified because it allows for the control of the generated effects. The SOM were able of capturing the diversity patterns generated for different quantities of migrants over different generations in various replicates. We concluded that SOMs are sensitive to detecting genetic variability and provide additional information on population organization.