QTL mapping of Poisson traits: a simulation study
Alex de Oliveira Ribeiro, Eduardo Bearzoti, and Thelma Sáfadi
QTL mapping consists of estimating the position and effects of genes (or linkage groups) that control quantitative traits. For traits that follow a normal distribution of probability, statistical approaches for QTL mapping are described in literature. However, for other traits the amount of research is little. The present work aimed to evaluate four alternative methods for QTL mapping of genes that control Poisson traits. The studied approaches were: standard interval mapping methodologies of Lander and Botstein and of Haley and Knott, a mixed model based on Poisson distributions, and a generalized linear model. Through computational simulation, the results showed great similarity for the different approaches under study. All approaches were highly effective for the tested simulations, especially when the abnormality of data in use was taken into account. Owing to the computational simplicity, Harley and Knott’s methodology was considered most effective.