This is a wrapper for the phylogenetic likelihood library (libpll) of the Exelixis lab.

Requirements:

• cython
• libpll installed from git - version 1.0.0 is too old

Installing libpll:

• on Mac PLL can be installed via homebrew using the homebrew-science tap (this should also work for linuxbrew)

  • brew tap homebrew/homebrew-science

  • brew install --HEAD libpll

• GCC compiler versions 5 and above may need the -fgnu89-inline flag to compile libpll's inline functions, which are written in a pre-C99 style.

pllpy has a single class, pll. This wraps the pllInstance type of libpll. Initialising a pll object can be done as follows:

import pllpy
instance = pllpy.pll(alignment_file, partitions, starting_tree, nthreads, random_seed)

• alignment_file is a Phylip-formatted input alignment file, passed as a string
• partitions is the RAxML-format partition list that associates a phylogenetic model (or models) to the alignment. This format is described in the RAxML manual - it's the format of the file passed by -q to RAxML. There is an example below. The value must be either a string filepath pointing to a file containing the partition description, or a string containing the partition description directly.
• starting_tree is one of three choices. If it is a string, then it must be a file path pointing to a newick-formatted tree file. Otherwise it should be True, to generate a parsimony tree, or False, to generate a random tree.
• nthreads sets the number of threads the pllInstance can use, if pllpy is linked to a multithreaded build of libpll.
• random_seed is passed to the libpll random number generator.

Maximum likelihood optimisation can be done using the pll.optimise() method:

# Optimise tree topology and all parameters, updating the topology every 5 steps (default):
instance.optimise(rates=True, branches=True, freqs=True, alphas=True,
                  topology=True, tree_search_interval=5, final_tree_search=True)
                  
# Equivalently:
instance.optimise()
                  
# Just optimise the model parameters, not the tree topology:
instance.optimise(topology=False)

# Print the log-likelihood
print(instance.get_likelihood())

The partition format is a list of regions of the alignment file:

Model, label1 = Start - End
Model, label2 = Start - End
Model, label3 = Start - End

Start - End defines the range of alignment columns. Model associates a phylogenetic model to the column range. This is GTR for DNA data, and DAYHOFF, DCMUT, JTT, MTREV, WAG, RTREV, CPREV, VT, BLOSUM62, MTMAM, LG, MTART, MTZOA, PMB, HIVB, HIVW, JTTDCMUT, FLU, STMTREV, LG4M, LG4X, GTR for protein data. AUTO can be used to automatically select the best model by AIC. Adding the suffix F to a model uses fixed empirical base frequencies. Adding X lets the base frequencies be optimised.