Identification of low frequency and rare variants for hypertension using sparse-data methods

Abstract

Availability of genomic sequence data provides opportunities to study the role of low-frequency and rare variants in the etiology of complex disease. In this study, we conduct association analyses of hypertension status in the cohort of 1943 unrelated Mexican Americans provided by Genetic Analysis Workshop 19, focusing on exonic variants in MAP4 on chromosome 3. Our primary interest is to compare the performance of standard and sparse-data approaches for single-variant tests and variant-collapsing tests for sets of rare and low-frequency variants. We analyze both the real and the simulated phenotypes.

Two‐phase designs for joint quantitative‐trait‐dependent and genotype‐dependent sampling in post‐GWAS regional sequencing

“…Interestingly, we identified one locus that is not in the list of "causal" variants, chr3:47734700, this can be explained by the LD structure in the MAP4 region (supplementary Fig. S6B) where we can observe that chr3:47734700 is in high LD with two of the causal variants: chr3:47956424 and chr3:47958037 (see Shin, Yi, and Bull, 2016, for details on this phenomenon).…”

Above and beyond state-of-the-art approaches to investigate sequence data: summary of methods and results from the population-based association group at the Genetic Analysis Workshop 19

“…The fifth trait was Q1 provided Alternative allele counts (NALTT field) were extracted with VCFtools and converted to 2-allele genotype calls. Nonbiallelic and monomorphic variants, and variants with more than 5 % missing calls were excluded, leaving 313,340 variants for analysis Shin et al [ 3 ] WES data in MAP4 from 1943 unrelated subjects Real data: Cases were defined as persons with SBP >140 mm Hg, DBP >90 mm Hg or taking antihypertension medication. Other persons, including individuals with a missing medication field, were treated as controls Excluded 92 individuals with missing phenotype data Predicted alternative allele counts (DOSAGE field) were extracted with VCFtools; monomorphic variants were filtered out, leaving 90 variants for analysis Simulated phenotypes: Null trait Q1 (dichotomomized) and PHEN, both with disease prevalence of 17.8 % Thompson and Fardo [ 4 ] Variants in TNN , LEPR , GSN , TCIRG1 , and FLT3 including 100,000 base pairs upstream and downstream Simulated phenotypes Q1 and PHEN on 1943 unrelated subjects Data extracted with VCFtools; monomorphic variants were filtered out Wang et al [ 8 ] WES data 5 kb within, up- and downstream of MAP4 from 1943 unrelated subjects Simulated data, including a null trait (25 variants have true SBP effects) Excluded 81 subjects without age information; monomorphic and low-coverage (<20×) variants were filtered out, leaving 94 variants DBP diastolic blood pressure, GWSNPA genome-wide single nucleotide polymorphism array, MAF minor allele frequency, NALTT number of nonreference alleles for each individual thresholded, SBP systolic blood pressure, VCF variant call format, WES whole exome sequence …”