The effect of mixed selected and unselected samples on the power of QTL mapping

It is known from previous work that selection by truncation can lead to substantial loss of power to detect linkage with a QTL We examine the effect of mixed selected and unselected samples. We show that even if a modest proportion of the sample originates from an unselected population, this loss of power is quite effectively neutralized. Moreover, we reach the unexpected conclusion that if the selection intensity in the selected subpopulation is high, the power to detect QTL may actually increase.