Background: A quantitative and a binary trait for the 14th QTLMAS 2010 workshop were simulated under a model which combined additive inheritance, epistasis and imprinting. This paper aimed to compare results submitted by the participants of the workshop.
Methods: The results were compared according to three criteria: the success rate (ratio of mapped QTL to the total number of simulated QTL), and the error rate (ratio of false positives to the number of reported positions), and mean distance between a true mapped QTL and the nearest submitted position.
Results: Seven groups submitted results for the quantitative trait and five for the binary trait. Among the 37 simulated QTL 17 remained undetected. Success rate ranged from 0.05 to 0.43, error rate was between 0.00 and 0.92, and the mean distance ranged from 0.26 to 0.77 Mb.
Conclusions: Our comparison shows that differences among methods used by the participants increases with the complexity of genetic architecture. It was particularly visible for the quantitative trait which was determined partly by non-additive QTL. Furthermore, an imprinted QTL with a large effect may remain undetected if the applied model tests only for Mendelian genes.