Large-scale Whole Genome Sequencing Study for Bone Mineral Density: the UK10K Consortium. H. Zheng1, V. Forgetta1, S. Wilson2,3, C. Greenwood1, N. Timpson4, N. Soranzo5, T. Spector3, B. Richards1,3 1) McGill University, Montreal, Quebec, Canada; 2) University of Western Australia, Western Australia, Australia; 3) Kings College London, London, UK; 4) Bristol University, Bristol, UK; 5) Sanger Institute, UK.
Aim: To identify genetic variants, including rare variants, associated with bone mineral density (BMD), we performed an association study using whole genome sequencing data within UK10K Consortium (http://www.uk10k.org/). Methods: 3781 samples from TwinsUK (discovery study) and ALSPAC were whole genome sequenced in Sanger Institute. The haplotypes of sequenced individuals were then imputed into an additional 3896 TwinsUK GWAS samples (replication study). We tested the association of TwinsUK samples at lumbar spine (L1-L4) (LS), femoral neck (FN) and forearm (Total radius and ulna) BMD separately in these two datasets. The phenotypes were standardized adjusting for age, age2 and weight. We undertook single variant tests, as well as a collapsing test of coding variants, with GEMMA and skatMeta, respectively. We then meta-analyzed the association results from the sequenced and imputed TwinsUK datasets. We set significant and suggestive thresholds for single variant test at 5x10-8, and 5x10-6, and set significance threshold for collapsing tests at 2.5x10-6 (0.05 over ~20,000 genes across the genome). Results: In the single variant test meta-analysis of data from the discovery and replication studies for forearm BMD, 143 variants from WNT16 locus were significant, and another 125 variants were suggestive. For LS and FN BMD, we identified 156 (including 11 LRP5 variants) and 125 suggestive variants. In the metaSKAT analysis of coding variants, we identified WNT16 and HSPB1 to be strongly associated with forearm BMD, however, when we excluded common variants (MAF>5%) in gene region, WNT16 was no longer significant, suggesting common missense variants in WNT16 drove the association signal. Nine rare coding variants in HSPB1 were collapsed and had an association P-value of 9.8x10-7, and this association was largely driven by rs28708645 (MAF= 0.005, P=9.4x10-8). We also identified 3 genes (including LRP5) and 11 genes for LS and FN BMD, respectively. Conclusion: In this whole genome sequencing data analysis for BMD, we identified a novel, rare missense variant in a non-canonical HSPB1 transcript (ENST00000447574) as an important genetic determinant of forearm BMD, and confirmed the association of variants in WNT16 and LRP5 with forearm and LS BMD, respectively, and highlighted other novel loci and genes. These variants will soon undergo additional replication in upcoming de-novo genotyping.
You may contact the first author (during and after the meeting) at