TEPAA (Threshold-based Efficient Pairwise Association Approach)
TEPAA is a strategy to reduce the number of SNP pairs needed for epistasis detection while maintaing similar power with brute force approach. We use TEPAA to compute which SNP pair should be checked to see whether the p-value is significant. The p-value is computed by other software package, such as Pylmm and EMMAX.
As explained in the paper(Citation: ), TEPAA has two stages. In the first stage, we compute p-value for all single SNPs. Only those SNPs pass a certain threshold will advance to second stage. In the second stage, we combine SNPs from first stage to make SNP pairs and compute p-value for each SNP pair.
The first stage thresholds is highly related to the Minor Allele Frequencies (MAFs) of the SNPs. So we propose different thresholds for SNP pairs with different MAFs.
By default, we split all SNPs into 10 bins according to their MAFs. For each pair of bins, we compute a pair of thresholds. And SNPs with in one bin share the same threshold. Those SNPs pass the threshold will be paired up for epistatsis detection.