Knowledge Center Catalog

Refining penalized ridge regression : a novel method for optimizing the regularization parameter in genomic prediction

Montesinos-Lopez, A.

Refining penalized ridge regression : a novel method for optimizing the regularization parameter in genomic prediction - Bethesda, MD (United States of America) : Oxford University Press, 2024.

Peer review Opén Access

The popularity of genomic selection as an efficient and cost-effective approach to estimate breeding values continues to increase, due in part to the significant saving in genotyping. Ridge regression is one ofthe most popular methods used for genomic prediction; however, its efficiency (in terms of prediction performance) depends on the appropriate tunning of the penalization parameter. In this paper we propose a novel, more efficient method to select the optimal penalization parameter for Ridge regression. We compared the proposed method with the conventional method to select the penalization parameter in 14 real data sets and we found that in 13 of these, the proposed method outperformed the conventional method and across data sets the gains in prediction accuracy in terms of Pearson's correlation was of 56.15%, with not-gains observed in terms of normalized mean square error. Finally, our results show evidence of the potential of the proposed method, and we encourage its adoption to improve the selection of candidate lines in the context of plant breeding.


Text in English

2160-1836 (Online)

https://doi.org/10.1093/g3journal/jkae246


Genomics
Plant breeding
Breeding Value
Marker-assisted selection
Best linear unbiased predictor
Statistical models

International Maize and Wheat Improvement Center (CIMMYT) © Copyright 2021.
Carretera México-Veracruz. Km. 45, El Batán, Texcoco, México, C.P. 56237.
If you have any question, please contact us at
CIMMYT-Knowledge-Center@cgiar.org