An R package for computing gene p-values using the Adaptive Joint Test. This package can be used to analyze genes and pathways based on a genetic association study with a binary case-control outcome.
Single-marker analysis that evaluates one genetic marker a time is the most commonly used approach for the identification of disease susceptibility loci in genome-wide association (GWA) studies. Increasing empirical evidence has suggested that there are regions or genes consisting of multiple genetic variants, which jointly contribute to the disease risk. A gene-based test, which evaluates the association between the outcome and all SNPs in the gene simultaneously, can be a more effective approach than the single-marker approach for studying such regions, thus is helpful to uncover some of missing heritability. The package contains 3 types of gene-based tests: AdaJoint, AdaJoint2, and ARTP. The AdaJoint test starts the greedy search with the best marginal model with one SNP, hence its performance partially depends on the power of marginal test. An alternative strategy is starting from the best model with two SNPs through an exhaustive search (AdaJoint2). ARTP is the Adaptive Rank Truncated Product test. The main function is adajoint which allows the user to pass in a data frame which contains all the data, or allows the data to be stored in files and have the function read in the data. If the user already has files containing test statistics for each gene, then the function pathway.pvals can be called to compute the gene and pathway p-values. This package also includes 4 gene databases: hg16, hg17, hg18 and hg19 which can be used with TPED genotype files.
Kai Yu and Han Zhang