Data for "Modeling site-specific mutation patterns in pandemic-scale phylogenetics"

Data for "Modeling Site-Specific Mutation Patterns in Pandemic-Scale Phylogenetics" This repository contains simulated data and analyses presented in the manuscript. This manuscript assesses the performance of a rate variation model we call the site-specific-matrix (SSM) model, and compares this to other models of rate variation. All data is provided as a single compressed tarball. Simulated Data Four sets of simulated data are included in the repository. Each set consists of 10 replicates of a simulated alignment in MAPLE format. Sets correspond to alignment lengths of 50k, 100k, 200k, and 500k. Alignments simulated using the commands found in scripts/run_phastSim_50k.batch and scripts/run_merge_50k.batch. Commands for the other lengths are identical with alignment length and filenames changed appropriately. Reference sequences and reference trees can be found in the references directory. Analysis Output from analyses are contained in the output directory. These are split by simulated data and real SARS-CoV-2 data. We analysed the data simulated under the SSM model as above, and included in the simulations directory. For each replicate we performed the following analyses: cmaple ( UNREST model, i.e. no rate variation model, in CMAPLE) cmaple-rate-variation (scalar rate variation in CMAPLE) cmaple-site-specific-matrices (SSM model in CMAPLE) cmaple-true-rates (the true rates are given to CMAPLE, no rate estimation) MAPLE (UNREST model, i.e. no rate variation model, in MAPLE) MAPLE-rate-variation (scalar rate variation in MAPLE) For 200k simulated alignments we performed an extra analysis where rates were additionally estimated after each subtree prune and regraft (SPR) search of the tree (cmaple-site-specific-matrices-SPR). Likelihood values and Robinsons-Foulds distances are collated and included in separate directories for plotting. For the real SARS-CoV-2 data, we performed two analyses in CMAPLE, one with the SSM model (simulations/Viridian_2M_SARS-COV-2/ssm) and one with the scalar rate variation model (simulations/Viridian_2M_SARS-COV-2/rv). The exact commands for the 50k datasets can be found in the scripts directory. The commands used for other simulated datasets can be obtained by changing 50k to the appropriate length. The commands for the real SARS-CoV-2 data can also be found in the scripts directory. Plots Code for all plots produced in the paper can be found in the Jupyter notebook plots/plots.ipynb. Code/software The rate variation models developed in this work have been implemented in CMAPLE v2, which can be found at https://github.com/iqtree/cmaple. The exact CMAPLE code used to run analyses presented in the manuscript can be found at https://github.com/SR-Martin/cmaple. The MAPLE code used to run analyses can be found at https://github.com/NicolaDM/MAPLE. The software phastSim used to simulate data can be found at https://github.com/NicolaDM/phastSim.