Division of Biomedical Statistics and Informatics, Department of Health Sciences Research, Mayo Clinic College of Medicine, Rochester, MN 55905, USA.
Pharmacogenomics J. 2012 Apr;12(2):105-10. doi: 10.1038/tpj.2010.83. Epub 2010 Nov 9.
Often, analysis for pharmacogenomic studies involving multiple drugs from the same class is completed by analyzing each drug individually for association with genomic variation. However, by completing the analysis of each drug individually, we may be losing valuable information. When studying multiple drugs from the same drug class, one may wish to determine genomic variation that explains the difference in response between individuals for the drug class, as opposed to each individual drug. Therefore, we have developed a multivariate model to assess whether genomic variation impacts a class of drugs. In addition to determine genomic effects that are similar for the drugs, we will also be able to determine genomic effects that differ between the drugs (that is, interaction). We will illustrate the utility of this multivariate model for cytotoxicity and genomic data collected on the Coriell Human Variation Panel for the class of anti-purine metabolites (6-mercaptopurine and 6-thioguanine).
通常，对于涉及同一类多种药物的药物基因组学研究的分析，是通过分别分析每种药物与基因组变异的关联来完成的。然而，通过分别分析每种药物，我们可能会失去有价值的信息。在研究同一药物类别的多种药物时，人们可能希望确定能够解释个体对药物类别反应差异的基因组变异，而不是每个个体药物。因此，我们开发了一个多变量模型来评估基因组变异是否会影响药物类别。除了确定对药物相似的基因组效应外，我们还能够确定药物之间存在差异的基因组效应（即相互作用）。我们将说明该多变量模型在收集有关细胞毒性和基因组数据的柯里尔人类变异面板上对嘌呤代谢物类（6-巯基嘌呤和 6-硫鸟嘌呤）的应用。
