Explore chapters and articles related to this topic
Common Statistical Issues in Ophthalmic Research
Published in Ching-Yu Cheng, Tien Yin Wong, Ophthalmic Epidemiology, 2022
It is not mandatory to always adjust for multiple testing. If the presented research work is clearly stated as exploratory and an acknowledgment is made that the research is being conducted in order to generate a hypothesis for further testing, then multiple test corrections may not be needed. It is best however to avoid multiple testing. Multiplicity is very commonly seen in genome-wide association studies where a very large number of genetic markers are tested. Here strict adjustment has become standard practice and a genome-wide significance level of 5 × 10–8 is often adopted. In bio-informatics studies more sophisticated multiple correction adjustments are made which allow for the non-independence of tests with specific software developed, such as Myriads, to cope with this issue [31].
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 GWAS are analyzed by taking the SNPs one at a time and performing a hypothesis test of association with a single trait or disease outcome, using an additive genetic model. This is similar to what is done in candidate-gene studies, and yet there are specific issues in GWAS. The most important is the problem of multiple testing; to guard against false positives, very stringent significance thresholds are used with the current consensus being that p < 5 × 10−8 is the appropriate level for declaring genome-wide significance. Such a low threshold means that very large samples are required to provide the power to detect the small genetic associations that are found with most traits. For instance, to have 90% power at genome-wide significance to detect an association between a binary trait with prevalence of 20% and a SNP that has a true odds ratio of 1.1 per allele and an allele frequency of 10%, we would require about 55,000 cases and 55,000 controls (calculation performed using QUANTO sample size calculator by USC Biostats, available at http://biostats.usc.edu/software).
Introductory Remarks
Published in Dongyou Liu, Handbook of Tumor Syndromes, 2020
GWAS typically examine a genome-wide set of genetic variants in two large groups of individuals to see if any variant is associated with a trait. As single-nucleotide polymorphisms (SNP) tend to occur more frequently in people with a particular disease than in people without the disease, they constitute a useful target for GWAS. A genome-wide significance level of 0.05 has been shown to correspond to a LOD score of 3.3 or higher. GWAS is powerful for detecting common risk variants which have small individual effects on risk, whereas linkage analysis is valuable for detecting genes with rare, high penetrance risk variants [14,15].
Increasing sample diversity in psychiatric genetics – Introducing a new cohort of patients with schizophrenia and controls from Vietnam – Results from a pilot study
Published in The World Journal of Biological Psychiatry, 2022
V. T. Nguyen, A. Braun, J. Kraft, T. M. T. Ta, G. M. Panagiotaropoulou, V. P. Nguyen, T. H. Nguyen, V. Trubetskoy, C. T. Le, T. T. H. Le, X. T. Pham, I. Heuser-Collier, N. H. Lam, K. Böge, I. M. Hahne, M. Bajbouj, M. M. Zierhut, E. Hahn, S. Ripke
Next, we assessed how much variance in schizophrenia risk could be attributed to underlying genetic variation. After quality control and genotype imputation (see Supplementary material), 343 individuals remained for polygenic risk score analysis. Schizophrenia polygenic risk scores (SCZ-PRS) were generated with GWAS summary statistics from East-Asian (EAS), European (EUR), and mixed-ancestry populations as training sets. In line with previous findings, individual-level SCZ-PRS were significantly associated with schizophrenia case status (Figure 1, Supplementary Table S1). However, the predictive accuracy of SCZ-PRS among Vietnamese participants varied with the ancestral composition of the training data sets. The strongest effect sizes were observed when PRS were derived from trans-ancestry summary statistics, which explained ∼4.9% of the variance in schizophrenia liability (p = 6.83 × 10−8 at Pd <1). Likewise, PRS constructed on East-Asian GWAS summary statistics (PGC SCZ wave 3) accounted for ∼4.5% of the variability in schizophrenia risk (p = 2.73 × 10−7 at Pd < 1). PRS trained on European ancestry GWAS results also predicted SCZ status in the Vietnamese target sample with an explained variance of only ∼3.9% (p = 1.79 × 10−6 at Pd < 0.01). As expected, best overall predictions were observed at more inclusive p-value thresholds that capture a larger proportion of the polygenic signal by including more independent common variants below the genome-wide significance level.
Atherosclerotic plaque injury-mediated murine thrombosis models: advantages and limitations
Published in Platelets, 2020
MFA Karel, B. Hechler, MJE Kuijpers, JMEM Cosemans
A different approach was employed by Baaten et al. [43], whom developed a novel synthesis method to quantitatively compare studies on the role of mouse genes in arterial thrombosis and overcome limitations caused by sample size and differences in methodology. Of the 431 studied mouse genes 60 genes showed a consistent effect on murine arterial thrombosis. For these 60 genes, an overall high homology on the nucleotide level was present with the human orthologs. Also, a network was constructed with human protein orthologs of 267 genes with modifying effects on murine arterial thrombosis. This network covered substantial gene sets identified in GWAS of stroke, CAD, platelet count and volume, and related studies. Each approach has its strengths and limitations. Understanding the model characteristics is of vital importance. For the study by Baaten et al. [43] holds among others that genes with a role in the vascular component of thrombotic disorders are underrepresented. However, GWAS only detect variants that are common in the population and whose effects on risk (odds ratio) are large enough to become significant at the very stringent genome-wide significance level (10−7–10−8). Statistical power is in turn strongly dependent on population size and, importantly, an association that arises from a GWAS does not necessarily imply a causal relationship [58]. Hence, studying candidate genes that emerge from human GWAS in mouse models could either confirm the prediction made GWAS and/or provide new (contrasting) information.
Genome-wide association study of white-coat effect in hypertensive patients
Published in Blood Pressure, 2019
Jenni M Rimpelä, Teemu Niiranen, Antti Jula, Ilkka H Pörsti, Antti Tikkakoski, Aki Havulinna, Terho Lehtimäki, Veikko Salomaa, Kimmo K Kontula, Timo P Hiltunen
The Q-Q plot from the discovery GWAS of systolic WCE showed little evidence of genomic inflation and supported the idea that some of the associations may be significant (Supplementary Figure 2(A)). Manhattan plot of the results (Figure 1(A)) illustrates four loci with p < 1 × 10−5, but no SNPs with genome-wide significance (p < 5 × 10−8) were identified. The four loci with p < 1 × 10−5 in the discovery GWAS are presented in Table 2, and the local Manhattan plots are depicted in Supplementary Figure 3(A–D). As the GENRES subjects received four different antihypertensive monotherapies in a rotational fashion separated by one-month placebo periods (Supplementary Figure 1), it was pertinent to test if the top associations derived from the placebo periods were similar when WCE data during administration of the four different drugs were analyzed. The effects of the SNPs were consistently in the same direction as during placebo periods and p values for the associations often remained statistically significant (Supplementary Table 2). In the GENRES study, the WCEs during the drug treatment periods did not differ statistically significantly from each other, nor from the mean WCE during placebo periods.