In this study, a QTL mapping method combining multiple traits is proposed. The key feature of the new method is that it tests all possible traits and chooses those affected traits in the model for a specific QTL, whereas these “working” traits may be different for different QTL. The BIC model selection criteria was used to choose the best model and to ascertain the “working” traits; then the “working” traits were combined together to detect QTL affecting these traits, by which it can boost QTL signals and increase statistical power. The new method has been applied to Barley dataset, which contains eight traits. The results showed that sometimes a QTL might affect more than one trait but not always affected all traits; the QTL detection power with new method was increased and 4 more QTLs were detected compared with single-trait method; generally, − logP values were much higher than those by single-trait analysis, but they were highly correlated, confirmed the unbiaseness of the new multitrait method. These results suggest that the developed multitrait method is effective for QTL mapping. The computational program is written in Fortran language for convenience to use, which is available on request.

在这项研究中，提出了一种结合了多个性状的QTL映射方法。新方法的关键特征是它可以测试所有可能的特征，并为特定的QTL选择模型中受影响的特征，而这些“有效”特征对于不同的QTL可能有所不同。BIC模型选择标准用于选择最佳模型并确定“工作”特征。然后将“工作”性状组合在一起以检测影响这些性状的QTL，从而增强QTL信号并提高统计能力。新方法已应用于包含八个特征的大麦数据集。结果表明，有时一个QTL可能影响多个特征，但并不总是影响所有特征。新方法的QTL检测能力比单性状法提高了4个。通常，− logP值比单性状分析的logP值高得多，但它们之间具有高度相关性，证实了新的多性状方法的不偏倚。这些结果表明，所开发的多特征方法对于QTL作图是有效的。为了方便使用，该计算程序以Fortran语言编写，可应要求提供。