Genome-wide association testing of Biolog GenIII phenotypes vs. accessory genome presence/absence

The association of accessory gene occurrence with phenotypic profiles obtained with the Biolog GenIII (continuous values) was tested using the phylogenetic framework implemented in the ‘treeWAS’ R package (Collins & Didelot 2018). The test was performed on 9,621 non-trivial OG presence/absence profiles using the ML method for ancestral state reconstruction. Significance was assessed by comparing test scores to those obtained by simulating random genotype profiles along the 10-taxon species tree for 100 times more loci than the tested dataset (ca. 105 simulations) and scaling the p-values using FDR correction. Phenotypes with significant hits were then tested again with more stringent parameters, using 1,000 times more loci than the tested dataset and Bonferroni p-value correction.

本研究借助‘treeWAS’ R包（Collins & Didelot 2018）所实现的系统发育分析框架，检验附属基因（accessory gene）的携带状态与通过Biolog GenIII平台获取的连续数值型表型谱之间的关联。本次检验基于9,621个非平凡同源基因组群（Orthologous Group, OG）的存在/缺失谱，采用用于祖先状态重建的最大似然法（Maximum Likelihood, ML）完成。显著性评估通过将检验得分与模拟得到的得分进行比较实现：沿包含10个类群的物种树模拟随机基因型谱，模拟的基因座数量为待检验数据集的100倍（约10^5次模拟），并采用错误发现率（False Discovery Rate, FDR）校正对p值进行校准。对于检出显著关联的表型，本研究进一步采用更为严格的参数重新开展检验：本次检验使用的基因座数量为待检验数据集的1000倍，并采用邦弗朗尼（Bonferroni）p值校正方法。