Initial Multi-Omics Model Analysis

Introduction

Methane emissions from ruminants significantly impact global climate, prompting the United Nations’ goal for a 30% reduction by 2030.

The FAO (2019) emphasizes balancing emission reduction with sustaining ruminant productivity for global food systems.

Traditional methods like dietary changes and feed additives face challenges, particularly with rumen microbiomes’ adaptability.

Genetic selection offers a sustainable approach to reducing emissions, leveraging the heritable nature of methane emission traits.

A major challenge is the low heritability of methane traits and difficulty in accurately measuring emissions on a large scale.

Incorporating omics data (metabolomics, genomics, proteomics, microbiomes) can refine breeding value estimation models.

Hayes (2017) and Weisharr et al. (2020) have integrated omics data into genetic models, enhancing the understanding of microbiomes’ impact on phenotypes.

Christensen et al. (2021) and Zhao et al. (2022) developed advanced methodologies, including non-linear interactions and missing observation solutions.

Microbiomes, though not direct intermediary traits, are crucial in linking genotype to methane production.

Our study evaluates a multiomics model incorporating microbiome data, using PCA to manage high-dimensional data and to improve predictive ability.

The study aims to demonstrate the effectiveness of integrating microbiome information into genomic prediction models, potentially enhancing genetic selection strategies for reducing methane emissions.

Dataset and Methodology

The dataset was divided into training and test groups, with 1,052 individuals encompassing the years 2014, 2015, and 2016.

Specifically, 691 individuals from 2014 and 2015 were allocated to the training group, while 361 individuals from 2016 formed the test population.

All individuals were genotyped using a high-density Melanpaper 600K SNP chip, further refined to 15K SNPs for focused analysis.

This approach allowed for a focused analysis on four traits: Methane, Methane Ratio, Live Weight, and Carbon Dioxide.

Genomic Prediction Accuracy and Bias Analysis

Analysis for Methane Trait

A comparison between 15K and 50K SNP chips using training populations from 2014 and 2015 and a test population from 2016.

Methane Trait Analysis

Analysis for Methane Ratio Trait

Methane Ratio Trait Analysis

Analysis for live weight

Live Weight Analysis

Analysis for carbondioxide

Carbon Dioxide Analysis

Next step is to compare accuracy and bias for the population split by year of birth versus five fold

Genomic Prediction Accuracy and Bias Analysis for Methane Trait: A Comparison Between five fold and split by year( Using Training Populations from 2014 and 2015 and a Test Population from 2016)

Genomic Prediction Accuracy and Bias Analysis for Methane ratio Trait: A Comparison Between five fold and split by year( Using Training Populations from 2014 and 2015 and a Test Population from 2016)

Genomic Prediction Accuracy and Bias Analysis for live weight: A Comparison Between five fold and split by year( Using Training Populations from 2014 and 2015 and a Test Population from 2016)

Genomic Prediction Accuracy and Bias Analysis for carbodioxide: A Comparison Between 15K and 50K SNP Chips Using Training Populations from 2014 and 2015 and a Test Population from 2016)