Comparison of Association Testing Processes

Examine the features and capabilities of the association testing processes (found in the GWAS Testing and Other Association Testing Genetics subcategories) in the table below.

Process	Tests for association between trait(s) and...	In Unrelated samples, Families, or Both1?	Taking into account	Using these methods	Can accommodate traits of type...			Tests marker effects and GxE interaction effects?
					Binary	Continuous	Other (nominal, ordinal, or survival)
Case-Control Association	Markers	Unrelated	Population structure (when applying VIF for genomic control)	Pearson chi-square tests, Fisher exact tests, C-A trend test	Y
PCA for Population Stratification	SNPs (coded numerically)	Unrelated	Population structure	EIGENSTRAT (Price et al. 2006)	Y	Y
SNP-Trait Association2 3	SNPs	Both	Fixed and random effect	Regression, ANOVA, and linear mixed models	Y	Y	Y	Y
Imputed SNP-Trait Association4	SNPs (represented by genotype probabilities)	Both	The uncertainty of called genotypes; fixed and random effects	Regression, ANOVA, and linear mixed models	Y	Y	Y	Y
Survey SNP-Trait Association	SNPs	Unrelated	Complex survey designs and very large data sets	Regression models and frequency table statistics for survey data	Y	Y		Y
Quantitative TDT	Markers	Families5	Population structure through use of families and covariates	Various quantitative transmission disequilibrium tests	Y	Y
TDT6	Markers	Families7	Population structure through use of families	Transmission/disequilibrium test and variations of it	Y
GWAS Meta-Analysis	SNPs, from multiple studies		Multiple GWASs	Meta-analysis methods: Stouffer et al. 1949; Wang and Bushman 1999
Multiple SNP-Trait Association	Numerically coded genotypes from multiple SNPs at a time	Both	Fixed and random effects	Regression models, kernel machine method, Hotelling's T-squared test, and Sequence Kernel Association Test (SKAT)8	Y		Y	Y
Rare Variant Tutorial	Rare variants		The value of guidance in choosing processes and parameters	Genetics Rare Variants Workflow, Rare Variant Association	See help on underlying processes.
Rare Variant Association	SNPs (rare variants and optionally common variants)	Unrelated	Multiple variants within a gene or pathway	VT, WSS, aSum-P, KBAC, CMAT, C-alpha, and others	Y	Y
Pleiotropic Association	SNPs	Unrelated	Testing for multiple traits at once	Regression, ANOVA, and MANOVA		Y		Y
Marker-Trait Association9	Markers	Both	Fixed and random effects	Regression, ANOVA, and linear mixed models	Y	Y		Y
SNP-SNP Interactions	Interactions between pairs of SNPs	Both	SNP main effects, other fixed and random effects	Regression, ANOVA, and linear mixed models	Y	Y	Y
SNP Interaction Discovery	Interactions between pairs of SNPs	Unrelated		Penalized Logistic Regression (PLR) (Park and Hastie 2008)	Y
Q-K Mixed Model2 10	SNPs	Both	Population structure and family relatedness (represented by Q and K matrices, respectively)	Regression, ANOVA, and linear mixed models (Yu et al. 2006)	Y	Y		Y

---

1

Relatedness can be accounted for in random effects.

2

Can handle multiallelic markers, but converts them to SNPs for analysis in terms of major allele.

3

Estimated SNP capacity per run: 1,000 to 1,000,000.

4

Unlike other genetic association processes, this process requires a stacked input data set.

5

Trios, siblings, or extended pedigrees with IBD data.

6

Facilitates the analysis of X-linked markers.

7

Trios and/or discordant sib-pairs.

8

Wu, M.C., S. Lee, et al. (2011). Rare-variant association testing for sequencing data with the sequence kernel association test. American Journal of Human Genetics 89:82-93.

9

Estimated SNP capacity per run: 1 to 100,000.

10

You must have already completed the Q and K matrices.

Return to GWAS Testing or Other Association Testing.

See Genetics for other subcategories.