Harnessing Phenotypic and Genomic Data to Reduce Lameness in Dairy Cattle

Summary: Aims 1. Assess machine learning variable selection in GWAS.

2. Apply the results from Objective 1, alongside traditional approaches, to GWAS of computer vision lameness-associated phenotypes. 3. Estimate and validate genetic parameters and breeding values for individual foot lesions.

4. Quantify the survival risk for foot lesions and incorporate these results into a whole-farm simulation model to estimate the environmental impact. Methodology

This project will initially use simulated data to assess and develop variable selection methods for GWAS. Subsequently, data from recent or ongoing projects will be analysed, specifically a dataset of around 5,000 genotyped cattle with computer vision phenotypes (TS/W022168/1), and a large dataset of foot lesion records collated from foot-trimmers (BB/X017451/1).

These data will support GWAS of novel computer vision lameness-associated traits, and the estimation and validation of foot lesion breeding values, calculated within a single-step framework. Foot lesion data will be merged with farm records, through collaboration with the industry partner, to quantify herd survival associated with each foot lesion, and to estimate the environmental cost of foot lesions by incorporating these results into a whole-farm simulation model.