Genetic variation in chromatin state across multiple tissues in Drosophila melanogaster

We use ATAC-seq to examine chromatin accessibility for four different tissues in Drosophila melanogaster: adult female brain, ovaries, and both wing and eye-antennal imaginal discs from males. Each tissue is assayed in eight different inbred strain genetic backgrounds, seven associated with a reference quality genome assembly. We develop a method for the quantile normalization of ATAC-seq fragments and test for differences in coverage among genotypes, tissues, and their interaction at 44099 peaks throughout the euchromatic genome. For the strains with reference quality genome assemblies, we correct ATAC-seq profiles for read mis-mapping due to nearby polymorphic structural variants (SVs). Comparing coverage among genotypes without accounting for SVs results in a highly elevated rate (55%) of identifying false positive differences in chromatin state between genotypes. After SV correction, we identify 1050, 30383, and 4508 regions whose peak heights are polymorphic among genotypes, among tissues, or exhibit genotype-by-tissue interactions, respectively. Finally, we identify 3988 candidate causative variants that explain at least 80% of the variance in chromatin state at nearby ATAC-seq peaks. Author summaryChromatin states are well described in Drosophila melanogaster embryos, but adult and pre-adult tissues are poorly studied, as are differences among genotypes. We carried out ATAC-seq on four different tissues in eight different inbred genotypes with biological replicates within tissue and genotype. We discover that apparent differences in coverage, and by inference chromatin openness, are often due to segregating structural variants (SVs) that can only be corrected for if strains are associated with high-quality genome assemblies. After correction for false positives associated with SVs, we identify thousands of regions that appear to vary in chromatin state between genotypes or vary between genotypes in a tissue-dependent manner. It has been widely speculated that cis-regulatory variants contribute to standing variation in complex traits. If this is true, chromatin states that vary between individuals, perhaps in a tissue-dependent manner, are likely to be enriched for quantitative trait loci.