20 December, 2020

Genomic Improvement In Dairy Cattle

Study design

Data was collected on 4,000 dairy cattle (2,800 adults) in 3 lactation classes at Grosvenor Farms, UK. Ear tissue samples were collected and low-coverage, whole genome sequence data was obtained on all individuals. A set of genetic variants was then generated (59 million) using standard variant calling methods. Additional subsets representing content equivalent to 50K, 100K and 777K SNP chips and WGS variant datasets were supplied. QC measures and biostatistical methods were applied to estimate several environmental and management factors as well as variance components for each trait. Linear mixed-effect modelling was applied on the datasets to account for differences caused by fixed and random effects.

The Challenge

Synomics’ combination of a proprietary technology platform and deep industry experience offers unique potential to identify trait-associated, combinatorial and predictive SNPs that standard GWAS does not find and translate these SNPs to breeding models which out-perform current best-in-class.

In this study, we used key commercial traits in dairy cattle (lameness, pregnancy success to first insemination, mastitis and 305-day fat yield) across multiple SNP densities (from 50k to 59m) to investigate the additional, detailed insight our platform was able to discover. The study sought to quantify the impact this additional insight has when translated to a genomic prediction model and genomic selection.

Our Solution

Genotype and phenotype datasets were analysed in Synomics’ proprietary platform, identifying high-order combinations of SNPs (including epistatic interactions). These combinations capture the non-linearity of biological effects and the impact they have on phenotypes much better than existing methods based on single features (such as GWAS). These novel, biologically, and statistically relevant genetic variants were then incorporated into our next-generation genomic evaluation method.

> 0 %
reduction in prediction error
> 0 %
increase in phenotype prediction accuracy
> 0 %
increase in genomically enhanced breeding value (gEBV) reliability
> 0 %
increase in heritability

The Results

Despite the comparatively small dataset, Synomics’ platform identified signals sufficient to detect 13 – 117 highly-predictive SNPs which mapped to between 6 – 31 genes, which are potential targets for intervention. We were able to find these SNPs despite small sample sizes and high SNP chip densities. Indeed, increasing SNP density (777k and 59 million low pass sequence) found more novel signals.  Conversion of these unique SNP combinations into a standard genomic prediction model showed step-change improvement in key metrics as well as a materially wider EBV distribution compared with standard breeding models.  This results in an increased genetic gain in dairy cattle breeding.


Synomics has licensed global use of this proprietary technology platform in bovine exclusively to Vytelle Holdings Inc.

Related Reports

Stack of cassava
October 3, 2022
Case Study

Genomic Prediction of Cassava Mosaic Disease

Golden retriever in the meadow
September 15, 2022

Synomics Announces Partnership with Mars Petcare

Cow in the meadow
March 21, 2022

Frontier In Genetics