An alternate F-test is applied. The F ratio is MSSub- Population/ MSStrain(Sub- Population). This test is used because the individual mice in the study are not independent, but rather are replicates within a strain. Therefore, we use the appropriate error term and degrees of freedom for the random effect of strain nested within the sub-populations. To account for multiple testing, the false discovery rate (FDR) was controlled using q-value estimation for the sex and strain main effects and sex × strain interaction effects. The R/q-value software developed by Storey (2002) was used for the analyses. A q-value threshold of 0.05 was used to identify significant results. Quantitative trait locus mapping was also performed separately for sub-populations.
