Genomic prediction of leaf rust resistance to Arabica coffee using machine learning algorithms

Abstract

Genomic selection (GS) emphasizes the simultaneous prediction of the genetic effects of thousands of scattered markers over the genome. Several statistical methodologies have been used in GS for the prediction of genetic merit. In general, such methodologies require certain assumptions about the data, such as the normality of the distribution of phenotypic values. To circumvent the non-normality of phenotypic values, the literature suggests the use of Bayesian Generalized Linear Regression (GBLASSO). Another alternative is the models based on machine learning, represented by methodologies such as Artificial Neural Networks (ANN), Decision Trees (DT) and related possible refinements such as Bagging, Random Forest and Boosting. This study aimed to use DT and its refinements for predicting resistance to orange rust in Arabica coffee. Additionally, DT and its refinements were used to identify the importance of markers related to the characteristic of interest. The results were compared with those from GBLASSO and ANN. Data on coffee rust resistance of 245 Arabica coffee plants genotyped for 137 markers were used. The DT refinements presented equal or inferior values of ApparentError Rate compared to those obtained by DT, GBLASSO, and ANN. Moreover, DT refinements were able to identify important markers for the characteristic of interest. Out of 14 of the most important markers analyzed in each methodology, 9.3 markers on average were in regions of quantitative trait loci (QTLs) related to resistance to disease listed in the literature.

Artificial Neural Networks in the Prediction of Genetic Merit to Flowering Traits in Bean Cultivars

“…González-Camacho et al [19] concluded that a Probabilistic Neural Network was more accurate than ANN-MLP for assigning maize and wheat lines. In addition, [14], considering the accuracy of the prediction of leaf rust resistance, showed that methodologies based on Computational Intelligence (including ANN) performs better than G-BLASSO.…”

“…The ANNs recognize patterns and regularities of data and represent an alternative as a universal approximation of complex functions [13]. In a GS context, this feature allows automatically to fit factors such as epistasis and dominance since it is not necessary to know a priori if the data have these effects [14]. In addition, this approach does not require any assumptions about the distribution of phenotypic values as the statistical methods do.…”

Machine learning and statistics to qualify environments through multi-traits in Coffea arabica

“…Each model is used as an individual classifier. A new individual will be allocated to the most common class among the predictions of the individual B classifiers [ 38 ]. The bagging estimate is defined by [ 2 ]: …”

“…The random forest results in a process of eliminating the correlation between the trees generated, further improving the accuracy of forecasts. However, it uses a smaller number of predictive traits in each division concerning Bagging [ 20 , 38 ]. This is achieved in the tree-growing process through random selection of the input variables [ 58 ].…”

“…Used as a classification method, ANNs have certain advantages, such as being non-parametric [ 36 ] and tolerant of data loss [ 37 ]. Just like ANNs, classification trees, and their refinements do not require assumptions about the model [ 38 ]. Also, classification trees are better classifiers than linear statistical methodologies, allowing for non-linearity of data and easier interpretation [ 39 ], as it provides information on which attributes are most important for prediction or classification [ 26 , 27 , 40 , 41 ].…”