Geno (Norway)

BackgroundRegions of genome-wide marker data may have differing influences on the evaluated traits. This can be reflected in the genomic models by assigning different weights to the markers, which can enhance the accuracy of genomic prediction. However, the standard multi-trait single-step genomic evaluation model can be computationally infeasible when the traits are allowed to have different marker weights.ResultsIn this study, we developed and implemented a multi-trait single-step single nucleotide polymorphism best linear unbiased prediction (SNPBLUP) model for large genomic data evaluations that allows for the use of precomputed trait-specific marker weights. The modifications to the standard single-step SNPBLUP model were minor and did not significantly increase the preprocessing workload. The model was tested using simulated data and marker weights precomputed using BayesA. Based on the results, memory requirements and computing time per iteration slightly increased compared to the standard single-step model without weights. Moreover, convergence of the model was slower when using marker weights, which resulted in longer total computing time. The use of marker weights, however, improved prediction accuracy.ConclusionsWe investigated a single-step SNPBLUP model that can be used to accommodate trait-specific marker weights. The marker-weighted single-step model improved prediction accuracy. The approach can be used for large genomic data evaluations using precomputed marker weights.

The energy sector is undergoing a major transformation in the composition of its energy matrix and the new trends detected in different scenarios have been published by institutions and/or organizations in the sector. This document contributes to the search for solutions to the serious problems of environmental pollution and humanity's economic inequality. To achieve this, it examines the reports of energy sources, and carbon dioxide emissions and performs a regression analysis up to the year 2100. The results indicate that a 25% increase in renewable energy is required, compared to renewable energy. that was required before the pandemic, to limit the increase in the temperature of the earth's surface to 1.5°C average, concerning the pre-industrial period. Unfortunately, the planet's energy matrix shows no signs of becoming sustainable and climate change and economic inequality continue to affect humanity.

Selection for polled in Norwegian Red cattle started 30 years ago and intensified the last 5 years. The SNP chip includes the (celtic) polled mutation, making selection very accurate for genotyped animals. However, ~85% of calves are ungenotyped and phenotypic horn status has been used to predict the polled genotype. This prediction has low accuracy (r2=0.69) which limits genetic progress. To improve predictions, a broad set of variables related to horns were extracted from the herd registration system. Data for ~134,000 genotyped animals was used to train and compare different machine learning methods for predicting polled genotype. Gradient boosting machines performed best, followed by random forests, deep neural networks, and generalized linear models. The best gradient boosting machine gave a large improvement in accuracy (r2=0.87). We have implemented this in the routine evaluations and expect increased accuracy of selection for ungenotyped animals and faster increase in polled allele frequency.

In pig breeding, the final product is a crossbred (CB) animal, while selection is performed at the purebred (PB) level using mainly PB data. However, incorporating CB data in genetic evaluations is expected to result in greater genetic progress at the CB level. Currently, there is no optimal way to include CB genotypes into the genomic relationship matrix. This is because, in single-step genomic BLUP, which is the most commonly used method, genomic and pedigree relationships must refer to the same base. This may not be the case when several breeds and CB are included. An alternative to overcome this issue may be to use a genomic relationship matrix (G matrix) that accounts for both linkage disequilibrium (LD) and linkage analysis (LA), called G. The objectives of this study were to further develop the G matrix approach to utilize both PB and CB genotypes simultaneously, to investigate its performance, and the general added value of including CB genotypes in genomic evaluations. Data were available on Dutch Landrace, Large White, and the F1 cross of those breeds. In total, 7 different G matrix compositions (PB alone, PB together, each PB with the CB, all genotypes across breeds, and G) were tested on 3 maternal traits: total number born (TNB), live born (LB), and gestation length (GL). Results show that G gave the greatest prediction accuracy of all the relationship matrices tested for PB prediction, but not for CB prediction. Including CB genotypes in general increased prediction accuracy for all breeds. However, in some cases, these increases in prediction accuracy were not significant (at < 0.05). To conclude, CB genotypes increased prediction accuracy for some of the traits and breeds, but not for all. The G matrix had significantly greater prediction accuracy in PB than the other G matrix with both PB and CB genotypes, except in one case. While for CB, the G matrix with genotypes across all breeds gave the greatest accuracy, though this was not significantly different from G. Computation time was high for G, and research will be needed to reduce its computational costs to make it feasible for use in routine evaluations. The main conclusion is that inclusion of CB genotypes is beneficial for both PB and CB animals.