Computational protocol: Polymorphisms in NF-κB Inhibitors and Risk of Epithelial Ovarian Cancer

Similar protocols

Protocol publication

[…] The selection of informative tagSNPs from among a larger pool of available SNPs allows for maximal genomic coverage and reduced genotyping redundancy []. We identified tagSNPs within five kb of NFKBIA (chromosome 14q13.2, RefSeq NM_020529.1) and NFKBIB (chromosome 19q13.2, RefSeq NM_002503.3) using the algorithm of ldSelect [] to bin pairwise-correlated SNPs at r2 ≥ 0.80 with minor allele frequency (MAF) ≥ 0.05 among publicly-available European-American data from the National Heart, Lung, and Blood Institute's Program for Genomic Applications SeattleSNPs gene-resequencing effort []. Within bins of SNPs in linkage disequilibrium (LD), tagSNPs with the maximum predicted likelihood of genotype success (Illumina-provided SNP_Score, San Diego, CA) were selected. Within each gene, we binned 26 SNPs resulting in 13 tagSNPs in NFKBIA and eight tagSNPs in NFKBIB; four singleton SNPs in NFKBIA and two singleton SNPs in NFKBIB failed conversion in development of the custom genotype panel and were excluded. The inclusion of additional SNPs with particular suspected functional relevance further increases coverage in a hypothesis-based manner at minimal increased cost; thus, we included all putative-functional SNPs (within 1 kb upstream, 5' UTR, 3' UTR, or non-synonymous) with MAF ≥ 0.05 identified in Ensembl version 34 and Illumina-provided SNP_Score > 0.6, resulting in one additional 3' UTR and three additional 5' upstream SNPs in NFKBIA. A total of 13 NFKBIA SNPs and six NFKBIB SNPs were genotyped (see Additional file ). [...] Distributions of demographic and clinical variables were compared across case status using chi-square tests and t-tests as appropriate. Individual SNP associations for ovarian cancer risk were assessed using logistic regression, in which odds ratios (ORs) and 95% confidence intervals (CIs) were estimated. Primary tests for associations assumed an ordinal (log-additive) effect with simple tests for trend, as well as separate comparisons of heterozygous and minor allele homozygous women to major allele homozygous women (referent) using a 2 degree-of-freedom (d.f.) test. In addition, we used a gene-centric principal components analysis to create orthogonal linear combinations of minor allele counts. The component linear combinations that accounted for at least 90% of the variability in the gene were included in a multivariable logistic regression model and simultaneously tested for gene-specific global significance using a likelihood ratio test. Haplotype frequencies were also estimated within each gene and a global haplotype score test of association between haplotypes and ovarian cancer risk was conducted at the gene level using a score test []. Individual haplotype tests compared each haplotype to all other haplotypes combined. NFKBIA rs3138050 was excluded from gene-level analyses due to failed genotyping in Duke University participants. All analyses were adjusted for age, race, region of residence, body mass index, hormone therapy use, oral contraceptive use, parity, and age at first birth. We used SAS (SAS Institute, Cary, NC, Version 8, 1999), Haplo.stats http://mayoresearch.mayo.edu/mayo/research/biostat/schaid.cfm, and S-Plus (Insightful Corp, Seattle, WA, Version 7.05, 2005) software systems. […]

Pipeline specifications

Software tools ldSelect, haplo.stats
Application GWAS
Diseases Neoplasms, Ovarian Neoplasms