ten Napel, J. | bioRxiv Science

bioRxiv2023

Accelerated computations for iterative-solver techniques in single-step BLUP models

In the last decade, a number of methods have been suggested to deal with large amounts of genetic data in genomic predictions. Yet, steadily growing population sizes and the suboptimal use of computational resources are pushing the practical application of these approaches to their limits. As an extension to the C/CUDA library miraculix, we have developed tailored solutions for the computation of genotype matrix multiplications which is a critical bottleneck in the empirical evaluation of many statistical models. We demonstrate the benefits of our solutions at the example of single-step models which make repeated use of this kind of multiplication. Targeting modern Nvidia(R) GPUs as well as a broad range of CPU architectures, our implementation significantly reduces the time required for the estimation of breeding values in large population sizes. miraculix is released under the Apache 2.0 license and is freely available at https://github.com/alexfreudenberg/miraculix.

bioinformatics↗

bioRxiv2021

Impact of genomic preselection on subsequent genetic evaluations with ssGBLUP - using real data from pigs

BackgroundEmpirically assessing the impact of preselection on subsequent genetic evaluations of preselected animals requires comparison of scenarios taking into account different approaches, including scenarios without preselection. However, preselection almost always takes place in animal breeding programs, so it is difficult to have a dataset without preselection. Hence most studies on preselection used simulated datasets, concluding that genomic estimated breeding values (GEBV) from subsequent single-step genomic best linear unbiased prediction (ssGBLUP) evaluations are unbiased. The aim of this study was to investigate the impact of genomic preselection (GPS) on accuracy and bias in subsequent ssGBLUP evaluations, using data from a commercial pig breeding program. MethodsWe used data on four pig production traits from one sire line and one dam line. The traits are average daily gain during performance testing, average daily gain throughout life, backfat thickness, and loin depth. As these traits had different weights in the breeding goals of the two lines, we analyzed the two lines separately. Per line, we had a reference GPS scenario which kept all available data, against which the next two scenarios were compared. We then implemented two other scenarios with additional layers of GPS by removing all animals without progeny either i) only in the validation generation, or ii) in all generations. We conducted subsequent ssGBLUP evaluations per GPS scenario, utilizing all the data remaining after implementing the GPS scenario. In computing accuracy and bias, we compared GEBV against progeny yield deviations of validation animals. ResultsResults for all traits in both lines showed marginal loss in accuracy due to the additional layers of GPS. Average accuracy across all GPS scenarios in both lines was 0.39, 0.47, 0.56, and 0.60 respectively for the four traits considered in this study. Bias was largely absent, and when present did not differ greatly among corresponding GPS scenarios. ConclusionAs preselection generally has the same effect in animal breeding programs, we concluded that impact of preselection is generally minimal on accuracy and bias in subsequent ssGBLUP evaluations of selection candidates in pigs and in other animal breeding programs.

genetics↗