International Journal of Engineering Inventions e-ISSN: 2278-7461, p-ISSN: 2319-6491 Volume 6, Issue 12 [December. 2017] PP: 43-46 www.ijeijournal.com Page | 1 Application of artificial neural networks in the simulation with genetic data Leila Maria Ferreira 1 , Janaína de Andrade Silva 2 , Gustavo César Sant’Ana 3 , Geraldo Magela de Almeida Cançado 4 , Aluízio Borém 5 , Juliano Lino Ferreira 6 1 PhD student in Agricultural Statistics and Experimentation, Federal University of Lavras, Brazil, 2 PhD in Computational Modeling, Federal University of Juiz de Fora, Brazil, 3 Post-Doctor in Bioinformatics, Agronomic Research for Development, France, 4 Senior Researcher at the Brazilian Agricultural Research Corporation (Embrapa) - Center for Genetic Engineering and Molecular Biology, Campinas-SP, Brazil, 5 Teacher of the Department of Plant Science, Federal University of Viçosa, Brazil, 6 Researcher at the Brazilian Agricultural Research Corporation (Embrapa Livestock South), Bagé-RS, Brazil Abstract: The objective of this work was the concept of applying artificial neural networks in the study of genetic data, in order to make the identification of the microsatellite markers for a particular species of plant to be analyzed more efficient. In this study, was used as an experimental model the data generated for 26 grapevine genotypes were divided into the following populations: Vitis vinifera; North American varieties; and interspecific hybrid of rootstocks. After the network training was carried out, an error rate of 0.0003460 was obtained, concluding that the network was able to learn according to the type of data used, even when these data are small. Keywords: Network, Genotype, Grapevine --------------------------------------------------------------------------------------------------------------------------------------- Date of Submission: 27-12-2017 Date of acceptance: 09-01-2018 --------------------------------------------------------------------------------------------------------------------------------------- I. INTRODUCTION From the molecular point of view, a genetic marker (or locus marker) serves to identify a site or region of a chromosome. An ideal genetic marker should present a series of attributes: high polymorphism, good reproducibility, detection of large numbers of unlinked loci, and simple inheritance. However, the simplicity and low costs of the method are determining factors for the routine application of a molecular marker. Molecular markers are facilitating studies of genetics, taxonomy and plant evolution, providing a significant advance in scientific knowledge. The main implications of this advance in knowledge are reflected in the power, precision and speed in the manipulation of genetic variability. Thus, plant breeding may benefit in various ways with the application of molecular markers. Recently, a series of statistical analyzes (Principal coordinate analysis - PCoA, Structure, UPGMA, Closest neighbor, Statistical summaries as content of polymorphic information - PIC, allelic diversity, probability of identity - PI and exclusion - PE, allele number, frequency of alleles, among others) has been routinely and widely applied in the evaluation of data obtained with molecular markers. With the advancement of the multidisciplinarity, new opportunities are created for the use of neural networks, including the evaluation of data obtained by molecular markers. Artificial neural networks have been gaining more and more visibility in the studies associated with genetic data, showing an efficient technique of analysis. Among the several works in this area are: Khan et al. (2001), Kan et al. (2004), Lancashire et al. (2010), Narayanan et al. (2004), Coppedè et al. (2010), Ornella and Tapia (2010), Genoud et al. (2009), Cajas et al. (2009), Camacho et al. (2012), Pandolfi et al. (2001), Khoshroo et al. (2014), Pan et al. (2013) and Silva et al. (2014). According to Crestan (2017) there are several definitions for neural networks, also called neurocomputers, connection networks or parallel distributed processors. "A neural network is a massively distributed parallel processor being made up of simple processing units, which have the natural propensity to store experimental knowledge and make it available for use" (Haykin, 1999). A neural network resembles the human brain in two respects:  Knowledge is acquired by the network based on its environment through a learning process;  Connecting forces between neurons, known as synaptic weights, are used to store the acquired knowledge. With a focus on artificial neural networks, one of the most fascinating research areas today is the simulation of a human being's cognitive abilities. Machines are designed to exhibit intelligent behavior, as if they were human reactions. The intelligence of the human being is the most advanced within the universe of living organisms and the place that welcomes this intelligence within the human body is the brain. The basic