Integrating extensive functional annotations and multiomics of cattle enhances climate resilience prediction and mapping.

To understand the biological function of genomic regions, vast molecular data have been generated to annotate mammalian genomes. However, how to effectively use such extensive information to improve the mapping and prediction of complex traits, including those that respond to climate change, remains unresolved. Here, we apply a Bayesian framework to estimate a Functional-And-Evolutionary Multi-trait Importance (FAEMI) score that combines extensive functional annotations to predict the probability that a variable genomic site causes variation in 16 complex traits of 103 K cattle. The functional annotations include information from the transcriptome, epigenome, and metabolome of cattle as well as genome constraints across species from multiple genome annotation consortia, covering 2.13 million molecular phenotypes from 24 tissues/cell types of 8,446 cattle worldwide. FAEMI analyses quantify the phenotypic importance of functional assays to guide future annotation efforts and reveal significant correlations between molecular functionality and genotype-to-phenotype associations. In new data of 45 K cattle with heat tolerance phenotypes, the FAEMI score demonstrates significant advantages in improving genomic prediction and mapping. The FAEMI score improved genomic prediction accuracy of multiple heat tolerance phenotypes by ~11%. A cellular stress-related locus, stress-associated endoplasmic reticulum protein family member 2 (SERP2), was identified as underlying heat tolerance, with the lead variant (rs383130643) associated with enhancer activity. Additionally, high FAEMI-ranking variants are significantly enriched in variants affecting beef cattle traits. Together, our work provides methods and resources to map informative variants genome-wide, enhancing our understanding of the biology behind thermal tolerance and helping breed resilient cattle in a hotter world.