China
Africa
Explore chapters and articles related to this topic
Family-Based Case-Control Approaches to Study the Role of Genetics
Published in Ørnulf Borgan, Norman E. Breslow, Nilanjan Chatterjee, Mitchell H. Gail, Alastair Scott, Christopher J. Wild, Handbook of Statistical Methods for Case-Control Studies, 2018
Clarice R. Weinberg, Min Shi, David M. Umbach
Another problem that can influence population-based approaches is cryptic relatedness, where cases and controls that are treated as independent observations in a logistic analysis may actually have shared ancestry and not be fully independent. Methods for adjusting for that problem are provided in Chapter 27. Studies based on nuclear families are less subject to that problem (Bennett and Curnow, 2001).
Genetics of chronic pain: crucial concepts in genetics and research tools to understand the molecular biology of pain and analgesia
Published in Peter R Wilson, Paul J Watson, Jennifer A Haythornthwaite, Troels S Jensen, Clinical Pain Management, 2008
Bradley E Aouizerat, Christine A Miaskowski
Linkage analysis examines the cosegregation of genetic markers (e.g. SNPs) within families whereas association is meant to provide information on the involvement of specific alleles in a trait of interest in a group of unrelated individuals.21 One potential weakness of association studies is the fact that cryptic relatedness or population substructure (e.g. self-reported ethnicity may not adequately capture different subpopulations that may self-identify as the same ethnicity and could result in chance differences in the proportions of such subpopulations between cases and controls) thereby confounding the results. Although association studies are susceptible to confounding due to differences in population substructure, they have several advantages over linkage analysis, including: the need to recruit multigenerational families where the contribution of family members to the analysis is difficult to ascertain a priori is obviated;the power to detect alleles with weak effects, and the ability to provide estimates of the relative magnitude of the effects of multiple alleles.
The Meta-Analysis of Genetic Studies
Published in Christopher H. Schmid, Theo Stijnen, Ian R. White, Handbook of Meta-Analysis, 2020
Cosetta Minelli, John Thompson
Most genetic variants will have no impact on the disease under study, so if a histogram were plotted of all of the p-values from a primary study or from a GWAS meta-analysis, it should take the form of a uniform distribution with a small spike close to zero due to the few true associations. If we do not see this pattern, then something is wrong. An alternative graphical representation of the full set of significance tests is the quantile-quantile (QQ) plot, in which the sorted test statistics are plotted against their expected values under the null hypothesis. Almost equivalent to this is a plot of –log10 of the sorted p-values against the expected values under a uniform distribution, as illustrated in Figure 17.3. In such a plot, each point represents a SNP and the majority of points will cluster around a straight line representing equality between observed and expected values. When the slope of the points, ignoring the tail caused by the few true associations, deviates from 1, then we have a problem, which needs to be investigated. The slope is usually referred to as inflation factor or lambda (λ). The most likely causes of a problem in a single study are population stratification, genotyping error, or cryptic relatedness (relatedness among participants not known to the investigator). Deciding on the threshold for concern is made difficult by the correlation between variants that lie close together on the genome. So most researchers use informal criteria, typically they hope for a slope below 1.05 and get seriously concerned if it goes above 1.1. The slope is a useful indicator of a problem, but some people go further and adjust p-values using genomic control (see Section 17.2.3) so as to force agreement between the observed and expected p-values. This is sometimes justified as being conservative but amounts to an admission that there are problems that could not be identified. Genomic control might be applied at the level of the primary study, but after that, any inflation in the meta-analysis p-values (λ > 1) would suggest heterogeneity. Some people have applied a second level of genomic control to the meta-analysis, but this is likely to result in an unnecessary loss of power. A much better approach would be to investigate the heterogeneity and to correct for it through the use of random-effects models when necessary (Thompson et al., 2011).
A genome-wide association study of early gamma-band response in a schizophrenia case–control sample
Published in The World Journal of Biological Psychiatry, 2018
Bettina Konte, Gregor Leicht, Ina Giegling, Oliver Pogarell, Susanne Karch, Annette M. Hartmann, Marion Friedl, Ulrich Hegerl, Dan Rujescu, Christoph Mulert
Quality control and association analyses were performed in PLINK (http://pngu.mgh.harvard.edu/∼purcell/plink/) (Purcell et al. 2007). We excluded individuals with high missingness (>0.01) and mismatch between reported and estimated gender. Markers with call rate <99% or non-random missingness patterns between cases and controls (P < 1e-6) were excluded. Moreover, markers with minor allele frequency <0.02 and HWE P < 1e-5 in the sample tested for auditory induced oscillations were excluded from the analysis. A set of autosomal markers excluding the extended MHC region and pruned for linkage disequilibrium (LD) (r2 = 0.02, window size = 1500, step size 150) was used for cryptic relatedness, heterozygosity deviation and population stratification analyses. Principal components of the study sample were derived with EIGENSTRAT (Price et al. 2006).
Thyrotropin receptor antibodies and a genetic hint in antithyroid drug-induced adverse drug reactions
Published in Expert Opinion on Drug Safety, 2018
Lin-Chau Chang, Chien-Ching Chang, Pei-Lung Chen, Shun-Huo Wang, Yi-Hsuan Chen, Yung-Hsin Tsai, Shyang-Rong Shih, Wei-Yih Chiu, Cathy Shen-Jang Fann, Wei-Shiung Yang, Tien-Chun Chang
Briefly, for each individual genotyped in the GWAS, we applied quality control filters to the dataset to remove SNPs that were missing in >1% of samples, were not autosomal, had an MAF of <1%, or showed significant deviation from the Hardy–Weinberg equilibrium in controls (P < 1 × 10−4). In addition, we removed SNPs for which there was a significant difference in genotype call rates in case and control data (P < 1 × 10−6). For sample filtering, we excluded arrays with generated genotypes for <95% of the loci. We then calculated heterozygosity rates, and excluded those samples with deviations greater than seven standard deviations from the mean. We took advantage of the heterozygosity of X-chromosome SNPs to verify the gender of the individuals from whom the samples were obtained. We accordingly found that there was no gender mismatch in our samples. We used PLINK version 1.9 software (http://pngu.mgh.harvard.edu/~purcell/plink/) to identify samples with genetic relatedness indicating that they were either from the same individual (or monozygotic twins) or from first-, second-, or third-degree relatives. These determinations were based on evidence for cryptic relatedness from identity-by-descent status (pi-hat cutoff of 0.125). After filtering, we retained for analysis 574,162 SNPs identified from 69 cases with cutaneous reactions and/or hepatotoxicity, and 560 GD controls.
Exome Array Analysis of Nuclear Lens Opacity
Published in Ophthalmic Epidemiology, 2018
Stephanie J. Loomis, Alison P. Klein, Kristine E. Lee, Fei Chen, Samantha Bomotti, Barbara Truitt, Sudha K. Iyengar, Ronald Klein, Barbara E. K. Klein, Priya Duggal
All samples had a call rate >98%. Individuals with sex inconsistencies (n = 15) and Mendelian errors (n = 2) or missing values for nuclear sclerosis, age, sex, or smoking at the visit prior to cataract surgery or censoring (n = 312) were excluded. Cryptic relatedness and unexpected duplicates were determined by identifying pairs of individuals with identical by descent sharing >20%, representing first and second-degree relatives. The individual from each relative pair with the lower quality score (n = 85) was excluded.