Parallel Likelihood Calculation for Phylogenetic Comparative Models: the SPLITT C++ Library

Abstract

15Phylogenetic comparative methods have been used to model trait evolution, to test 16 selection versus neutral hypotheses, to estimate optimal trait-values, and to quantify the 17 rate of adaptation towards these optima. Several authors have proposed algorithms 18 calculating the likelihood for trait evolution models, such as the Ornstein-Uhlenbeck (OU) 19 1 . CC-BY-NC-ND 4.0 International license peer-reviewed) is the author/funder. It is made available under a The copyright holder for this preprint (which was not . http://dx.doi.org/10.1101/235739 doi: bioRxiv preprint first posted online Dec. 18, 2017; process, in time proportional to the number of tips in the tree. Combined with 20 gradient-based optimization, these algorithms enable maximum likelihood (ML) inference 21 within seconds, even for trees exceeding 10,000 tips. Despite its useful statistical 22 properties, ML has been criticised for being a point estimator prone to getting stuck in principle, the strategy of (i) and (ii) can be applied to any likelihood calculation on a tree 30 which proceeds in a pruning-like fashion, leading to enormous speed improvements. We 31 implement several variants of the parallel algorithm in the form of a generic C++ library, 32"SPLiTTree", capable to choose automatically the optimal algorithm for a given task and 33 computing platform. We give examples of models of discrete and continuous trait evolution

A Phylogeny-aware GWAS Framework to Correct for Heritable Pathogen Effects on Infectious Disease Traits

“…Our method relies on the freely available R package POUMM ( Mitov and Stadler 2017 ), which scales to trees of up to 10,000 tips ( Mitov and Stadler 2019 ). All code for the simulations and HIV spVL analysis presented in this study is available on the project GitHub at https://github.com/cevo-public/POUMM-GWAS .…”

Automatic generation of evolutionary hypotheses using mixed Gaussian phylogenetic models

“…The likelihood calculation was implemented in the R-package PCMBase (https://github.com/venelin/PCMBase), using internal calls to its Rcpp companion PCMBaseCpp (https://github.com/venelin/PCMBaseCpp) (24) and the SPLITT C++ library (https://github.com/venelin/SPLITT) (37). The RCP algorithm was implemented in the R-package PCMFit (https://github.com/venelin/PCMFit) (38).…”

Fast Bayesian inference of phylogenies from multiple continuous characters

“…6 in Mitov et al 2020). Readers can find a detailed description of this algorithm in Mitov et al (2020), with it being put to use in Mitov and Stadler (2019). We provide a worked example in the supplementary material.…”

“…Our BEAST2 implementation was extensively benchmarked and validated; it is correct, fast, scales linearly with the number of species, and supports both the inference of among-trait correlations with MCMC, as well as the use of the linear shrinkage method. Because it follows a general mathematical framework (Mitov and Stadler, 2019), our method can be readily extended to include models such as BM with trends (Hansen and Martins, 1996), Ornstein-Uhlenbeck (OU; Hansen and Martins, 1996;O'Meara et al, 2006), early burst (EB; Harmon et al, 2010), accelerated or decelerated-rate models (ACDC; Blomberg et al, 2003), to name a few. The BEAST2 platform also provides a series of clock and other evolutionary models that can be used in conjunction with (or replace components of) the integrative model we use here.…”