An organism-wide ATAC-seq peak catalog for the bovine and its use to identify regulatory variants

Abstract

We herein report the generation of an organism-wide catalogue of 976,813cis-acting regulatory elements for the bovine detected by the Assay for Transposase Accessible Chromatin using sequencing (ATAC-Seq). We regroup these regulatory elements in 16 components by nonnegative matrix factorization. Correlations between the genome-wide density of peaks and transcription start sites, between peak accessibility and expression of neighboring genes, and enrichment in transcription factor binding motifs supports their regulatory potential. Using a previously established catalogue of 12,736,643 variants, we show that the proportion of single nucleotide polymorphisms mapping to ATAC-seq peaks is higher than expected and that this is due to an ~ 1.3-fold higher mutation rate within than outside peaks. Their site frequency spectrum indicates that variants in ATAC-seq peaks are subject to purifying selection. We generate eQTL datasets for liver and blood and show that variants that drive eQTL fall into liver and blood-specific ATAC-seq peaks more often than expected by chance. We combine ATAC-seq and eQTL data to estimate that the proportion of regulatory variants mapping to ATAC-seq peaks is approximately 1 in 3, and that the proportion of variants mapping to ATAC-seq peaks that are regulatory is approximately 1 in 25. We discuss the implication of these findings on the utility of ATAC-seq information to improve the accuracy of genomic selection.

Sequenced-based GWAS for linear classification traits in Belgian Blue beef cattle reveals new coding variants in genes regulating body size in mammals

“…Beside these candidate coding variants, we found evidence for regulatory variants in three QTLR on BTA5, 18 and 23. CS from these three QTLR did overlap with the catalogue of regulatory regions identified by ATAC-SEQ by Yuan et al [ 54 ]. For QTLR on BTA5 and 23, there was also association between the CS with cis-eQTL from a study conducted in blood and liver in Holstein [ 54 ] and evidence for regulatory effects in the cattle GTEx dataset [ 55 ].…”

“…For the QTLR without obvious candidate coding variants, we also investigated whether SNPs in the CS overlapped with core and consensus segments from ATAC-seq peaks present in the catalogue from Yuan et al [ 54 ], with cis-eQTL reported in the same study or eQTL from the cattle GTEx study [ 55 ]. Contrary to the lead variant on BTA6, the CS variants on BTA5, consisting in six intronic variants from CCND2 (see Additional file 1 : Fig.…”

“…Contrary to the lead variant on BTA6, the CS variants on BTA5, consisting in six intronic variants from CCND2 (see Additional file 1 : Fig. S13), fall in consensus ATAC-seq peaks and even in the CS from an eQTL affecting CCND2 expression in liver [ 54 ] (the lead SNP matches both criteria, the alternate allele being associated with increased expression and higher height). Another SNP from the CS affects expression of TIGAR in muscle (see Additional file 3 : Table S23) [ 55 ].…”

“…Finally, in the CS for the QTLR on BTA23 containing only six SNPs (see Additional file 1 : Fig. S15), four intergenic variants match core ATAC-seq peaks and are in the CS from a blood eQTL reducing ARMC12 expression levels [ 54 ]. The lead SNP located upstream from ARMC12 is the lead SNP of this eQTL, the alternate allele is associated with lower expression and lower height.…”

“…Furthermore, information available from the literature was considered for variants previously reported. Finally, we investigated whether SNPs in the CS overlapped with core and consensus segments called from ATAC-seq peaks in a recently released catalogue [ 54 ], or with CS of blood and liver cis-expression QTL (eQTL) reported in the same study. This eQTL study was selected as we had access to all the data.…”