GLS model, but the magnitude of the difference was very small (Supplementary Fig. 3a). However, the running times for EMMA were substantially longer. Because the original EMMA re-estimates the variance parameters at each marker, given the size of the NFBC data set, it took more than 10 min of CPU time per marker on an Intel Xeon 3-GHz processor, even with an efficient C implementation of EMMA. A simple extrapolation suggests that it would take more than 6 years of CPU time to analyze a single GWAS data set using EMMA, taking a full mixed model approach. The total computational time using EMMAX for this data was 6.6 h in a single CPU, and the procedure could easily be parallelized to speed it up further.
