论文部分内容阅读
遗传模型的选择在基因-疾病关联研究Meta分析中是较为关键的难点。本文对Minelli等提出的无遗传模型约束的贝叶斯Meta分析方法进行深入讨论,利用JAGS和R语言对其进行实现,并在实例中比较了不同参数先验信息对合并效应量的影响,特别是外部临床先验的影响。实例研究结果显示,无论是无信息先验还是外部临床先验,在纳入研究个数较多的情况下,对合并效应量的影响均较小。
The choice of genetic model is a key challenge in Meta-analysis of gene-disease association studies. In this paper, we discuss the Bayesian Meta analysis method proposed by Minelli et al. Without any genetic model constraints and implement it by using JAGS and R language. In the example, we compare the influence of prior information of different parameters on the amount of the merger effect, especially It is the impact of external clinical priors. The results of the case study show that, no matter whether there is no prior information or external clinical prior experience, the impact of the combined effect is small when the number of included studies is large.