Graphical models are a powerful tool in modelling and analysing complex biological associations in high-dimensional data. The R-package netgwas implements the recent methodological development on copula graphical models to (i) construct linkage maps, (ii) infer linkage disequilibrium networks from genotype data, and (iii) detect high-dimensional genotype-phenotype networks. The netgwas learns the structure of networks from ordinal data and mixed ordinal-and-continuous data. Here, we apply the functionality in netgwas to various multivariate example datasets taken from the literature to demonstrate the kind of insight that can be obtained from the package. We show that our package offers a more realistic association analysis than the classical approaches, as it discriminates between direct and induced correlations by adjusting for the effect of all other variables while performing pairwise associations. This feature controls for spurious interactions between variables that can arise from conventional approaches in a biological sense. The netgwas package uses a parallelization strategy on multi-core processors to speed-up computations. The netgwas package is freely available at https://cran.r-project.org/web/packages/netgwas

翻译：图模型是建模和分析高维数据中复杂生物关联的有力工具。R软件包netgwas实现了copula图模型的最新方法论进展，用于：(i)构建连锁图谱，(ii)从基因型数据推断连锁不平衡网络，以及(iii)检测高维基因型-表型网络。netgwas可从有序数据以及混合有序-连续数据中学习网络结构。本文通过文献中的多个多元示例数据集应用netgwas的功能，展示了该软件包可获得的洞见。研究表明，与经典方法相比，该软件包提供了更现实的关联分析，因为它在执行成对关联时通过调整所有其他变量的影响来区分直接相关和诱导相关。这一特性在生物学意义上控制了传统方法可能产生的变量间虚假交互作用。netgwas软件包采用多核处理器并行化策略加速计算。该软件包可在https://cran.r-project.org/web/packages/netgwas免费获取。