Demographic inference under a spatially-continuous coalescent model

Simulation code used for data generation in Demographic inference under a spatially-continuous coalescent model.

Dependencies

• Python v2.7
• ERCS v1.0.1: https://pypi.python.org/pypi/ercs
• Discsim v1.0.0: https://pypi.python.org/pypi/discsim
• Arlequin v3.5.1.3: http://cmpg.unibe.ch/software/arlequin35/
• Seq-Gen v1.3.3: http://tree.bio.ed.ac.uk/software/seqgen/

Overview and Setup

The main simulation code is run from simulate.py. Parameter settings are specified in settings.py.

python simulate.py

The script runs simulations as follows:

1. Generate gene trees from Discsim
2. Simulate the corresponding DNA sequences with Seq-Gen
3. Convert the DNA sequences into the appropriate format for Arlequin

Simulation parameters are specified in settings.py. Output sequences are retained use to use with other analytical methods.

The simulation code sets up files for analysis in Arlequin, however Arlequin is not called directly. Instead Arlequin is run as a standalone through their provided scripts. For detailed instruction see the Arlequin manual.

The simulation code expects the compiled program Seq-Gen (named "seq-gen") in the seqgen folder. First download and compile the program from the link above. Rename the compiled program "seq-gen" and place in the folder named "seqgen."

Simulation parameters and random seeds used to run approximate Bayesian computation analysis are saved in the parameters.txt file. The seed used to run the simulation correspond to the names of the files generated for Arlequin.

Settings

Simulation settings a specified in the settings.py file.

Discsim Parameters

• length: specifies the size of the torus
• dimensions: specifies either 1D or 2D simulations
• sample_locations: coordinates for sample locations
• recombination_probability: probability of recombination along loci
• num_parents: set the number of parents of each sample

General Parameters

• num_replicates: specifies the number of simulations to run
• posterior_predictive_checks: if true, simulation parameters are drawn randomly from the file specified under "posterior_parameter_file"
• posterior_parameter_file: parameter values of retained simulations from ABC (see example file)
• neighborhood_size: the upper and lower bound for the neighborhood size prior. Values for neighborhood size are drawn from a uniform distribution.
• small_event: upper and lower bounds for the event prior.
• num_cpus: number of cores available to run simulations

SeqGen Parameters

• mutation_rates: the per locus mutation rate
• seq_length: sum of the number of base pairs of all loci
• num_partitions: total number of loci
• partitions: base pair length of each loci

Testing Parameters

• debug: set to False when running simulations