TreeFlow: Probabilistic programming and automatic differentiation for phylogenetics

Probabilistic programming frameworks are powerful tools for statistical modelling and inference. They are not immediately generalisable to phylogenetic problems due to the particular computational properties of the phylogenetic tree object. TreeFlow is a software library for probabilistic programming and automatic differentiation with phylogenetic trees. It implements inference algorithms for phylogenetic tree times and model parameters, given a tree topology. We demonstrate how TreeFlow can be used to quickly implement and assess new models. We also show that it provides reasonable performance for gradient-based inference algorithms compared to specialized computational libraries for phylogenetics., Carnivores sequence alignment accessed from benchmark in BEAST examples H3N2 sequence alignment taken from Vaughan TG, Kühnert D, Popinga A, Welch D, Drummond AJ. Efficient Bayesian inference under the structured coalescent. Bioinformatics. 2014 Aug 15;30(16):2272-9. doi: 10.1093/bioinformatics/btu201 , Data processing pipeline can be found at https://github.com/christiaanjs/treeflow-paper Tree topologies inferred using RAxML 8.2.12 Tree topologies are rooted using LSD 0.2 BEAST analyses are performed using BEAST 2.6.7 Variational inference analyses are performed using TreeFlow 0.0.1beta Sequences have been removed H3N2 BEAST XML as a result of license conflicts. This complete version of this file is generated by the above pipeline.概率编程（probabilistic programming）框架是开展统计建模与统计推断的强有力工具。但由于系统发育树（phylogenetic tree）对象的特殊计算特性，这类框架无法直接推广应用于系统发育（phylogenetic）相关问题。TreeFlow是一款面向概率编程与系统发育树自动微分（automatic differentiation）的软件库。在给定树拓扑结构（tree topology）的前提下，该库可实现针对系统发育树分化时间与模型参数的推断算法。本研究演示了如何利用TreeFlow快速实现并评估新型模型；同时相较于专门面向系统发育学的计算库，该工具在基于梯度的推断算法上表现出了较为合理的性能。 食肉动物序列比对数据取自BEAST示例中的基准测试集。 H3N2序列比对数据取自以下文献：Vaughan TG、Kühnert D、Popinga A、Welch D、Drummond AJ.《结构化溯祖模型下的高效贝叶斯推断》，发表于《生物信息学》，2014年8月15日；30(16):2272-2279，DOI: 10.1093/bioinformatics/btu201 数据处理流程可通过以下链接获取：https://github.com/christiaanjs/treeflow-paper 树拓扑结构通过RAxML 8.2.12软件推断得到。 树拓扑结构通过LSD 0.2软件完成定根。 BEAST分析通过BEAST 2.6.7版本执行。 变分推断（variational inference）分析通过TreeFlow 0.0.1beta版本执行。 因许可证冲突，H3N2 BEAST XML文件中的相关序列已被移除。本文件的完整版本系通过上述数据处理流程生成。