Association Analysis

Scenarios under Consideration [top]

Available statistics for WISARD depends on the presence of population stratification and the study design, and the statistical analysis usually starts from building the relationship matrix. Therefore they are summarized as follows:

• Relationship matrix
Relationship matrix is utilized for detecting the presence of population stratification and if there exists population stratification or family-based samples are utilized, relationship matrix should be incorporated to the statistical analysis as a correlation matrix. Therefore, builing relationship matrix is recommended as a first step of genetic association analysis.
• Population stratification
Results from genetic association analysis is invalid under the presence of population stratification, and it can be detected by using MDS plot. In addition, covariates which reveals the genetic similarity between individuals can be extracted with WISARD and if they are included as covariates for logistic/linear regression, the statistical analysis becomes robust againt the presence of population stratification.
• Association analysis
Statistics for genetic association analysis are illustrated for dichotomous and quantitative phenotypes under the absence of population stratification.
• Association analysis under the presence of population stratification
Statistics for genetic association analysis are illustrated for dichotomous and quantitative phenotypes under the presence of population stratification.
• Family-based association analysis
Statistics for family-based association analysis are illustrated for dichotomous and quantitative phenotypes.
• Epistasis analysis
• Gene-level analysis
Statistics for gene-level analysis are illustrated for dichotomous and quantitative phenotypes under the absence of population stratification.
• Gene-level analysis under the presence of population stratification
Statistics for gene-level analysis are illustrated for dichotomous and quantitative phenotypes under the presence of population stratification.
• Family-based gene-level analysis
Statistics for family-based gene-level analysis are illustrated for dichotomous and quantitative phenotypes.

Descriptive Measures Common for All Scenarios [top]

WISARD provides various statistics for the genetic association analysis with population-based samples. Statistics in WISARD assumes that each variant is biallelic and if there are variants with monomorphic or more-than-two alleles, they are excluded from the analysis.

If there exists some individuals with indel (insertion/deletion), their genotypes are excluded from the analysis by default but statistics which test the heterogeneity of indel between cases and controls can be calculated by using --indel option.

The following information for all statistics is always provided:

• chromosome number which the variant belongs to
• the physical position of each variant
• name of variant
• minor allele of each variant