We calculated the linkage disequilibrium between all pairs of variants in the Drosophila Genome Reference Panel with minor allele count ≥5. We used r2 ≥ 0.5 as the cutoff for a highly correlated SNP. We make available the list of all highly correlated SNPs for use in association studies. Seventy-six percent of variant SNPs are highly correlated with at least one other SNP, and the mean number of highly correlated SNPs per variant over the whole genome is 83.9. Disequilibrium between distant SNPs is also common when minor allele frequency (MAF) is low: 37% of SNPs with MAF < 0.1 are highly correlated with SNPs more than 100 kb distant. Although SNPs within regions with polymorphic inversions are highly correlated with somewhat larger numbers of SNPs, and these correlated SNPs are on average farther away, the probability that a SNP in such regions is highly correlated with at least one other SNP is very similar to SNPs outside inversions. Previous karyotyping of the DGRP lines has been inconsistent, and we used LD and genotype to investigate these discrepancies. When previous studies agreed on inversion karyotype, our analysis was almost perfectly concordant with those assignments. In discordant cases, and for inversion heterozygotes, our results suggest errors in two previous analyses or discordance between genotype and karyotype. Heterozygosities of chromosome arms are, in many cases, surprisingly highly correlated, suggesting strong epsistatic selection during the inbreeding and maintenance of the DGRP lines. LD205results.zipList of all SNP pairs in Freeze 2 of the Drosophila Genome Reference Panel with linkage disequilbirium r^2>0.5. Minor allele count of the focal SNP must be >=5.HouleMarquezSASfilesThis zip archive contains: Gcorrlimits.SAS. A SAS file that calculates limits on which range of minor allele frequencies can be correlated at r^2>=0.5. CalcHighCorr.SAS A SAS file that calculates the genotypic r^2 between sites that could be correlated at a particular r^2 value. gcorrexampledata.sas7bdat - as SAS format data set with sample data for CalcHighCorr.SAS.HouleMarquezF3_PCscoresSupplemental File S3. Inversion-typing of DGRP lines for the three common inversions In(2L)t, In(2R)NS, and In(3R)Mo, and heterozygosity of chromosome regions. For each inversion: the column beginning PC1sc is the score on Principal component 1 for variants between inversion breakpoints and in disequilibrium with large number of distant sites; (see text for explanation) . Pred is the inversion type predicted based on the PC1 scores (1=homozygous inversion, 0.5 Standard/inversion heterozygote, 0= homozygous Standard). CDL is the combined assignments of inversion-type from Langley et al. 2012 and Corbett-Detig & Hartl 2012 (1=homozygous inversion, 0= not a homozygote for the inversion, blank=not scored). Huang indicates the assignements from Huang et al. 2014 (0= homozygous Standard, 0.5= homozygous ST/INV, 1=homozygous inversion), blank= not scored), Heterozygosity columns are the average heterozygosity for chromsome segments defined by inversion breakpoints, plus the X and arm 3L. Heterozygosities were calculated from sites with no more than 5 missing calls in the 205 lines, and omit regions within 10kb of inversion breakpoints.