Metagenomic predictions for enteric methane emissions in sheep using long-read sequencing of rumen fluid samples

Metagenomic data were generated using Oxford Nanopore Technologies (ONT) platform for 396 sheep rumen fluid samples. Taxonomic and gene functional abundance matrices were constructed to develop predictive models for estimating methane emission phenotypes based on methane concentration from portable accumulation chambers (PAC). The greatest microbiability (𝑚2; proportion of phenotypic variation explained by the microbiome relationship matrices) was generated using Clusters of Orthologous Groups (COGs) functional annotations of the reads (𝑚2 = 0.92). Correlations between the predicted phenotypes and the PAC phenotypes in a 5-fold crossvalidation ranged from 0.48 to 0.51. These results indicate that metagenomic predictions could be implemented as a proxy methane phenotype in sheep and could be used to increase the feasible reference population size in genomic predictions for methane using multi-trait models.