For the shared features of all three tools (e.g. managing sequence datasets and association analyses of unrelated samples), we compared the computational efficiency for data preparation and association analyses using the 1000 Genomes Project Phase 3 dataset. This dataset consists of 2504 individuals genotyped on 84.8 million markers. We simulated continuous phenotypes based on the null model of no genotype–phenotype associations. To prepare sequence data for association analyses, VAT and PSEQ require building databases, while RVTESTS requires annotating sequence variants. VAT and PSEQ took 495 and 36 CPU hours, respectively, on a desktop with Intel® Xeon® CPU E5-2650 v2 @ 2.60 GHz to prepare and import data into SQLite databases. As a comparison, annotating the whole-genome datasets required only 17 h for RVTESTS. It is clear that RVTESTS offers considerable improvement in time complexity.
