Functional annotation and Bayesian fine-mapping reveals candidate genes for important agronomic traits in Holstein bulls

Jicai Jiang,Yang Da,Ellen Freebern,John B Cole,Paul M Vanraden,Li Ma

doi:10.1038/s42003-019-0454-y

Jicai Jiang, Yang Da + Show 4 more

Open Access

https://doi.org/10.1038/s42003-019-0454-y

Copy DOI

Abstract

A hundred years of data collection in dairy cattle can facilitate powerful studies of complex traits. Cattle GWAS have identified many associated genomic regions. With increasing numbers of cattle sequenced, fine-mapping of causal variants is becoming possible. Here we imputed selected sequence variants to 27,214 Holstein bulls that have highly reliable phenotypes for 35 production, reproduction, and body conformation traits. We performed single-marker scans for the 35 traits and multi-trait tests of the three trait groups, revealing 282 candidate QTL for fine-mapping. We developed a Bayesian Fine-MAPping approach (BFMAP) to integrate fine-mapping with functional enrichment analysis. Our fine-mapping identified 69 promising candidate genes, including ABCC9, VPS13B, MGST1, SCD, MKL1, CSN1S1 for production, CHEK2, GC, KALRN for reproduction, and TMTC2, ARRDC3, ZNF613, CCND2, FGF6 for conformation traits. Collectively, these results demonstrated the utility of BFMAP, identified candidate genes, and enhanced our understanding of the genetic basis of cattle complex traits.

Highlights

A hundred years of data collection in dairy cattle can facilitate powerful studies of complex traits
We imputed over 3 million selected sequence variants to 27,214 Holstein bulls after quality control edits, using the 1000 Bull Genomes data as reference
While a proportion of these newly discovered QTLs were identified to be associated with new traits, these results demonstrated the superior power of our GWAS in dairy cattle

Summary

Introduction

A hundred years of data collection in dairy cattle can facilitate powerful studies of complex traits. GCTA-COJO is capable of fast conditional analysis for fine-mapping in cattle[16], the use of summary statistics and LD data from a reference population can be suboptimal when direct genotype and phenotype data are available To address these problems, we develop a fast Bayesian Fine-MAPping method (BFMAP) that can efficiently integrate functional annotations with fine-mapping. The high LD in the cattle genome makes fine-mapping and functional enrichment studies difficult, the large sample size and improved methods can help identify candidate genes of complex traits as well as biologically informative enrichment of candidate variants in functional annotation data. The functional data enriched with variants associated with complex dairy traits will be useful for future cattle GWAS and genomic prediction studies. The initial model search results can be reused for estimating enrichment of causal effects of dairy traits for additional functional annotations that are being generated by the FAANG and related projects in cattle[18]

Methods

Results

Conclusion