Phylo-SMC

Usage overview

After following the installation instructions, the following scripts can be used (see below for more detailed instructions):

• ./run-synthetic.bash: to generate data and run SMC and MCMC on synthetic data
• ./run-rna.bash: to load RNA data from a file and run SMC and MCMC
• ./create-figures.bash: to create the figures in the Bouchard et al. 2011 paper from the raw data

run-synthetic.bash

Type ./run-synthetic.bash to see the usage and an example. Limitation: note that in this alpha release, only the priorprior kernel with coalescent prior and K2P likelihood model are available. This will be relaxed in future releases. After running one instance of this script (or of run-rna.bash), the output is available in the subdirectory ouput/x.exec, in which you can find the following files:

• results/results.csv : summary statistics of the results of that run, in CSV format. See first row for name of fields: Method is either MB (for MCMC), or LAZYSMC (for SMC), IterScale is the number of particles/iterations, Repeat can be ignored, Metric is the metric used, and value is its value for that experiment.
• output/* : a copy of the (potentially synthetic) dataset, and synthetic tree (if any) used in this run
• options.map : the command line options used for this run
• log : the full stdout output for this run
• inferred-trees/* : all the consensus trees produced by the systems tested in this run

run-rna.bash

Type ./run-rna.bash to see the usage and an example. Note that the input file should be a fasta file with aligned RNA sequences. The path should be either absolute or relative to the resources folder (not to run-rna.bash). The limitation described in run-synthetic applies to this command as well. After running an instance, of this script the same output as run-synthetic is available in the subdirectory ouput/x.exec, (see above).

create-figures.bash

Type ./create-figures.bash to create the figures from the data in the raw-data folder, resources/state/remote. The output is written in the graphics folder.

run-rna-simple.bash

Similar to run-rna, but instead of computing the consensus, dumps the weighted samples in a file called particles-dump.txt in the exec directory (output/[x].exec/particles-dump.txt). The format is, for each line, the weight of the particle, a tab, and then the tree in newick format corresponding to the particle. This command does not run Mrbayes either, to keep its requirements minimal.