University of Oxford
5 files

Summary statistics for mineral metabolism GWAS

Download all (1.2 GB)
posted on 2023-08-08, 12:34 authored by Fadil Hannan, Anibha Mahajan

Summary statistics for mineral metabolism GWAS in UK Biobank.

Genome-wide association studies of albumin-adjusted serum calcium and phosphate concentrations (Fadil Hannan, Anubha Mahajan, personal communication):

Serum albumin-adjusted calcium concentrations for UK Biobank participants were derived using the following equation: adjusted calcium (mmol/L)=total Calcium(mmol/L) + 0.0177 *(46.3 albumin (g/L)). Data from participants with eGFR(MDRD)<60ml/min/1.73m2 and 25-OH vitamin D concentrations <30nmol/L were excluded from association analyses for both serum phosphate concentrations and serum albumin-adjusted calcium concentrations. Analyses were performed using genotyped and imputed variants from the UK Biobank. Genotyping was undertaken using UK-BiLEVE and UK-Biobank Axiom Arrays, as described in the “Methods” section of the main text. Phenotypes were inverse normalised with additional adjustments for array, age, and sex. Analyses were undertaken in individuals of European ancestry using BOLT-LMM to account for population sub-structure and cryptic relatedness. Imputed SNPs of minor allele frequency (MAF) <1% and of imputation quality score <0.3 were excluded from analyses. Lead SNPs were identified from unconditional meta-analyses and loci defined as ±500kb surroundingeach SNP. Overlapping loci were merged as one locus. GCTA was used to perform a stepwise model selection procedure to select independently-associated SNPs within each 1Mb region of significance p<5x10-9. Directly genotyped variants underwent stringent quality control checks, including call rate per array, manual cluster plot checks and status in Gnomad. Only variants with MAF <1% and the coding or loss-of-function annotations of “missense variant”, “stop gain”, “frameshift variant”, “splice acceptor variant”, “splice donor variant”, “splice region variant”, “start lost”, or “stop lost”  were included. A significance threshold of p<5x10-6 was used to identify directly genotyped SNPs with significant associations with each phenotype.


Usage metrics

    University of Oxford


    No categories selected



    Ref. manager