When analyzing ∼6000 unrelated individuals at ∼100 000 markers, RMW/RAREMETAL provides a speed improvement of ∼600-fold compared with SCORESEQ/MASS (Tang and Lin, 2013). This difference in speed increases with sample size and number of studies. The R package metaSKAT (Lee et al., 2013) provides comparably fast computations, but variable threshold test is not provided. An important difference between RAREMETAL and these published tools is the ability to use linear mixed models to account for sample relatedness and/or population structure. Even when using linear mixed models to account for relatedness and population structure, RMW can handle large datasets. A mixed model analysis of 10 000 individuals at 238 000 markers used 6.1 h and 2 GB memory. With 12 GB memory, RMW was able to analyze 23 000 individuals in <5 days. Other features in RAREMETAL unique to other published tools are the flexibility of changing gene definitions and grouping strategies after individual studies have been analyzed and the ability to perform conditional meta-analysis.
