Gillespie simulations of advantageous drivers & deleterious passengers in cancer

This simulator is written in C with an emphasis on very-fast simulations of moderately-sized tumors. A forward Gillespie algorithm (the Next Reaction) is used to model cell birth, death, and mutations. This code is written at a low-level and not designed for ease-of-use for the general population.

The simulator is best accessed via a python wrapper. To download the code & build the python wrapper:

$ git clone https://github.com/cd-mcfarland/pdSim.git # get repository
$ cd pdSim
$ ./setup.py build_ext  # Build the python wrapper that runs the simulator

The simulator has a number of precompiler directives that alter the kind of simulation that is run, including:

1. The Distribution of Fitness Effects (DFEs) that driver & passenger mutations are drawn from,
2. Whether-or-not the entire lineage history of the dividing cells is tracked,
3. The epistasis effect between multiple mutations,
4. Whether-or-not Circulating Tumor Cells (CTCs) are generated (to be used to simulate tumor growth at distant stroma),
5. Whether-or-not drivers are drawn from a psuedo-random DFE, or an array of specific genes.

load_cells.py organizes these precompiler options into keyword arguments of a python function that then compiles the code accordingly and then returns a python class sim (all at runtime). sim then executes a simulation upon instantiation. The various outputs of the simulation are accessed via methods and attributes of the returned object. An example of usage:

from load_cells import cells
sim = cells(passenger_distribution=2, tree=1) # passengers are drawn from a Log-Normal DFE; lineage is tracked
tumor = sim(sd=0.25, N_0=300)                 # simulate tumor growth with non-default values of sd, N_0
passenger_mutations = tumor.fixed_mutations[tumor.fixed_mutations <= 0]  
# fixed_mutations is a numpy.ndarray that records the fitness effects of every mutation that sweeps to fixation
# in temporal order. Passengers are, by definition, neutral or deleterious mutations. 
print('Mean fitness cost of fixated passengers: {:.3f}'.format(passenger_mutations.mean()))
# Mean fitness cost of fixated passengers: -0.001
tumor.plot(figname='population_size_versus_time.pdf')

The doc strings of the various methods and attributes should clarify the kinds of output you can retrieve. However, in general the code is not very accessible and was never written for general use. Knowledge of C & cython is needed, if you want to extend the code.

simple.py also provides a useful, well-documented example of how the simulator might be used to model dN/dS statistics.

Please contact Christopher McFarland for all questions.