Skip to main content
ARS Home » Northeast Area » Ithaca, New York » Robert W. Holley Center for Agriculture & Health » Plant, Soil and Nutrition Research » Research » Publications at this Location » Publication #297130

Title: The effects of relatedness and GxE interaction on prediction accuracies in genomic selection: a study in cassava

Author
item LY, DELPHINE - Montpellier Supagro – International Center For High Education In Agricultural Sciences
item HAMBLIN, MARTHA - Cornell University
item RABBI, ISMAIL - International Institute For Tropical Agriculture
item GEDLI, MELAKU - International Institute For Tropical Agriculture
item BAKARE, MOSHOOD - International Institute For Tropical Agriculture
item GAUCH, HUGH - Cornell University
item OKECHUKWU, RICHARDSON - International Institute For Tropical Agriculture
item DIXON, ALFRED - International Institute For Tropical Agriculture
item KULAKOW, PETER - International Institute For Tropical Agriculture
item Jannink, Jean-Luc

Submitted to: Crop Science
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 1/7/2013
Publication Date: 5/3/2013
Citation: Ly, D., Hamblin, M., Rabbi, I., Gedli, M., Bakare, M., Gauch, H., Okechukwu, R., Dixon, A., Kulakow, P., Jannink, J. 2013. The effects of relatedness and GxE interaction on prediction accuracies in genomic selection: a study in cassava. Crop Science. 53(4):1312-1325.

Interpretive Summary: Genomic selection may accelerate breeding cycles by enabling prediction of future performance based on the DNA marker scores of selection candidates. Prior to selection, it is important to know how accurate predictions will be. Cross validation estimates this prediction accuracy by removing a subset of breeding lines, making predictions using information from the remaining lines, then evaluating accuracy on the subset. Accuracies assessed in this way may be inflated by high genetic relationship within the breeding lines or by genotype by environment interaction. We tested cross validation sampling schemes preventing these sources of inflation. Using these cross validation schemes reduced estimated accuracy between 0.00 and 0.07 for genetic relatedness and between 0.01 and 0.18 for genotype-environment interaction. These were small but consistent effects, showing that our new methods should be useful in generating unbiased estimates of future prediction accuracy. Our research also showed that genomic selection has the potential to accelerate gains in cassava.

Technical Abstract: Prior to implementation of genomic selection, an evaluation of the potential accuracy of prediction can be obtained by cross validation. In this procedure, a population with both phenotypes and genotypes is split into training and validation sets. The prediction model is fitted using the training set, and its accuracy is calculated on the validation set. The degree of genetic relatedness between the training and validation sets may influence the expected accuracy, and so may the genotype by environment interaction in those sets. We developed a method to assess these effects and tested it in cassava. We used historical phenotypic data available from the International Institute of Tropical Agriculture Genetic Gain trial and performed genotyping by sequencing for these clones. We tested cross validation sampling schemes preventing the training and validation sets from sharing a) genetically close clones or b) similar evaluation locations. Over 19 traits, plot basis heritabilities ranged from 0.04 to 0.66. The correlation between predicted and observed phenotypes ranged from 0.15 to 0.47. Across traits, predicting for less related clones decreased accuracy from 0 to 0.07, a small but consistent effect. For 17 traits, predicting for different locations decreased accuracy between 0.01 and 0.18. We conclude that genomic selection has potential to accelerate gains in cassava and the existing training population should give a reasonable estimate of future prediction accuracies.