ABSTRACT: Quantile Random Forest (QRF) is a non-parametric methodology that combines the advantages of Random Forest (RF) and Quantile Regression (QR). Specifically, this approach can explore non-linear functions, determining the probability distribution of a response variable and extracting information from different quantiles instead of just predicting the mean. This evaluated the performance of the QRF in the genomic prediction for complex traits (epistasis and dominance). In addition, compare the accuracies obtained with those derived from the G-BLUP. The simulation created an F2 population with 1,000 individuals and genotyped for 4,010 SNP markers. Besides, twelve traits were simulated from a model considering additive and non-additive effects, QTL (Quantitative trait loci) numbers ranging from eight to 120, and heritability of 0.3, 0.5, or 0.8. For training and validation, the 5-fold cross-validation approach was used. For each fold, the accuracies of all the proposed models were calculated: QRF in five different quantiles and three G-BLUP models (additive effect, additive and epistatic effects, additive and dominant effects). Finally, the predictive performance of these methodologies was compared. In all scenarios, the QRF accuracies were equal to or greater than the methodologies evaluated and proved to be an alternative tool to predict genetic values in complex traits.

RESUMO: Quantile Random Forest (QRF) é uma metodologia não paramétrica, que combina as vantagens do Random Forest (RF) e da Regressão Quantílica (QR). Especificamente, essa abordagem pode explorar funções não lineares, determinando a distribuição de probabilidade de uma variável resposta e extraindo informações de diferentes quantis em vez de apenas prever a média. O objetivo deste trabalho foi avaliar o desempenho do QRF em predizer o valor genético genômico para características com arquitetura genética não aditiva (epistasia e dominância). Adicionalmente, as acurácias obtidas foram comparadas com aquelas advindas do G-BLUP. A simulação criou uma população F2 com 1.000 indivíduos genotipados para 4.010 marcadores SNP. Além disso, doze características foram simuladas a partir de um modelo considerando efeitos aditivos e não aditivos, com número de QTL (Quantitative trait loci) variando de oito a 120 e herdabilidade de 0,3, 0,5 ou 0,8. Para treinamento e validação foi usada a abordagem da validação cruzada 5-fold. Para cada um dos folds foram calculadas as acurácias de todos os modelos propostos: QRF em cinco quantis diferentes e três modelos do G-BLUP (com efeito aditivo, aditivo e epistático, aditivo e dominante). Por fim, o desempenho preditivo dessas metodologias foi comparado. Em todos os cenários, as acurácias do QRF foram iguais ou superiores às metodologias avaliadas e mostrou ser uma ferramenta alternativa para predizer valores genéticos em características complexas.

ABSTRACT. The genotype by environment interaction is the main factor that influences the response of evaluated genotypes in trials of value for cultivation and use. Adaptability and stability analyses are fundamental to understanding the performance of genotypes in a growing region. Some of these methodologies incorporate previous information for recommending an extra group of genotypes denominated as specific ideotypes under certain cultivation conditions. Based on this strategy, the centroid method and its modifications have been widely used due to the simplicity of classification of the evaluated genotypes. However, these methodologies present problems in identifying adaptability patterns of some genotypes. Artificial intelligence techniques, such as fuzzy C-means, can be an alternative to reduce these difficulties, since they use, in addition to distance information between genotypes, memberships (measures quantifying how much an observation belongs to a particular class) to increase discriminatory power. Therefore, our aim was to propose and evaluate the phenotypic adaptability method by fuzzy clustering to assist cultivar recommendations. The adaptation of the fuzzy C-Means method to classify the genotypes was implemented in BioFuzzy software. The grain yield data of black common bean genotypes were used to evaluate the potential of the method. The results obtained by this method were compared with those obtained by the centroid method. The phenotypic adaptability method by fuzzy clustering was effective in identifying the adaptability patterns of common bean genotypes. Moreover, the discriminatory power was higher than that observed with the centroid method. ABSTRACT use region conditions strategy However techniques Cmeans, Cmeans C means, means C-means difficulties measures class Therefore recommendations CMeans Means software Moreover

ABSTRACT. This paper aimed to evaluate the effectiveness of subset selection of markers for genome-enabled prediction of genetic values using radial basis function neural networks (RBFNN). To this end, an F1 population derived from the hybridization of divergent parents with 500 individuals genotyped with 1000 SNP-type markers was simulated. Phenotypic traits were determined by adopting three different gene action models - additive, additive-dominant, and epistatic, representing two dominance situations: partial and complete with quantitative traits having a heritability (h2) of 30 and 60%; traits were controlled by 50 loci, considering two alleles per locus. Twelve different scenarios were represented in the simulation. The stepwise regression was used before the prediction methods. The reliability and the root mean square error were used for estimation using a fivefold cross-validation scheme. Overall, dimensionality reduction improved the reliability values for all scenarios, specifically with h2 =30 the reliability value from 0.03 to 0.59 using RBFNN and from 0.10 to 0.57 with RR-BLUP in the scenario with additive effects. In the additive dominant scenario, the reliability values changed from 0.12 to 0.59 using RBFNN and from 0.12 to 0.58 with RR-BLUP, and in the epistasis scenarios, the reliability values changed from 0.07 to 0.50 using RBFNN and from 0.06 to 0.47 with RR-BLUP. The results showed that the use of stepwise regression before the use of these techniques led to an improvement in the accuracy of prediction of the genetic value and, mainly, to a large reduction of the root mean square error in addition to facilitating processing and analysis time due to a reduction in dimensionality.

Resumo: O objetivo deste trabalho foi avaliar o método gráfico de rede de similaridade na classificação de genótipos de arroz (Orzya sativa) irrigado quanto a sua adaptabilidade e estabilidade. Duas medidas estatísticas foram utilizadas para representar a proximidade de comportamento (baseada na correlação de Pearson) ou de valores (baseada na distância de Gower) entre pares de genótipos ou entre genótipo e ambiente. Foram avaliados dados de produtividade de 18 genótipos de arroz irrigado em três locais de Minas Gerais, nas safras de 2012/2013, 2013/2014, 2014/2015 e 2015/2016, em delineamento em blocos ao acaso. Os genótipos foram previamente avaliados quanto à adaptabilidade e à estabilidade pelos métodos de Eberhart & Russell e centroide. As representações gráficas fornecidas pelas redes de similaridade facilitaram o reconhecimento do padrão de interação genótipo x ambiente, o que permitiu superar as dificuldades de interpretação devido à discordância entre os resultados obtidos com os métodos de Eberhart & Russell e centroide. As redes de similaridade facilitam os estudos de interação genótipo x ambiente.

Abstract: The objective of this work was to evaluate the similarity network graphic methodology for the classification of flood-irrigated rice (Orzya sativa) genotypes regarding their adaptability and stability. Two statistical measures were used to represent the proximity of the behavior (based on Pearson’s correlation) or values (based on Gower’s distance) between pairs of genotypes or between genotype and environment. Productivity data of 18 genotypes were evaluated in three locations in the state of Minas Gerais, Brazil, in the harvests of 2012/2013, 2013/2014, 2014/2015, and 2015/2016, in a randomized complete block design. The genotypes were previously assessed for adaptability and stability by the Eberhart & Russell and centroid methods. The graphical representations provided by the similarity networks allowed to better identify the pattern of the genotype x environment interaction, overcoming the interpretation difficulties due to the disagreements between the results obtained by the Eberhart & Russell and centroid methods. The similarity networks improve genotype x environment interaction studies.

ABSTRACT. The objective of this study was to compare different graphical dispersion analysis techniques in two- or three-dimensional planes. In this study, the data from different published works were used in order to determine the best methodology for analyzing the genetic diversity of different species. In this study, efficiency is measured by the amount of original distance absorbed by the projection of distances technique, which in the case of major components is equal to the amount of total variation originally available and retained by the principal components used for dispersion purposes. The projection of dissimilarity measurement technique, principal component analysis (PCA), and principal coordinate analysis (PCoA) were used. Considering the analysis by means of three orthogonal axes, the graphical dispersion efficiency was 82.22 for PCA, 87.22 for PCoA, and 85.25 for the projection of distances technique. For the 2D analysis, considering the two main axes, the mean dispersion efficiency was 69.90 for the PCA, 75.06 for the projection technique, and 78.16 for PCoA. Considering the studies carried out with experimental data of six different species, it is concluded that the principal coordinate analysis is superior.

Resumo: O objetivo deste trabalho foi avaliar o uso de redes neurais artificiais em comparação à modelagem por meio de modelos lineares generalizados na predição de resistência à ferrugem em café arábica (Coffea arabica). Foram utilizados 245 indivíduos provenientes de uma população F2, oriundos da autofecundação do híbrido F1 H511-1, resultante do cruzamento da cultivar suscetível Catuaí Amarelo IAC 64 (UFV 2148-57) e do genitor resistente Híbrido de Timor (UFV 443-03). Os 245 indivíduos foram genotipados com 137 marcadores. Realizaram-se análises com redes neurais artificiais e com modelos lineares generalizados sob o enfoque bayesiano. As redes neurais identificaram quatro marcadores importantes pertencentes a grupos de ligação que foram recentemente mapeados, enquanto o modelo generalizado bayesiano identificou somente dois marcadores pertencentes a esses grupos. Foram observadas taxas de erro de predição inferiores (1,60%) para predizer a resistência à ferrugem em café arábica, quando foram utilizadas as redes neurais artificiais em vez de modelos lineares generalizados sob o enfoque bayesiano (2,4%). Os resultados mostraram que as redes neurais artificiais são uma abordagem promissora para predizer a resistência à ferrugem em café arábica.

Abstract: The objective of this work was to evaluate the use of artificial neural networks in comparison with Bayesian generalized linear regression to predict leaf rust resistance in Arabica coffee (Coffea arabica). This study used 245 individuals of a F2 population derived from the self-fertilization of the F1 H511-1 hybrid, resulting from a crossing between the susceptible cultivar Catuaí Amarelo IAC 64 (UFV 2148-57) and the resistant parent Híbrido de Timor (UFV 443-03). The 245 individuals were genotyped with 137 markers. Artificial neural networks and Bayesian generalized linear regression analyses were performed. The artificial neural networks were able to identify four important markers belonging to linkage groups that have been recently mapped, while the Bayesian generalized model identified only two markers belonging to these groups. Lower prediction error rates (1.60%) were observed for predicting leaf rust resistance in Arabica coffee when artificial neural networks were used instead of Bayesian generalized linear regression (2.4%). The results showed that artificial neural networks are a promising approach for predicting leaf rust resistance in Arabica coffee.