ts-ipc: inter-process communication with tree sequences

ts-ipc provides a light-weight framework for end-to-end population genetic simulations that utilize tree sequence recording to combine forward simulation and coalescent simulation.

It facilitates communication between the forward simulator SLiM and python, where much of the tree sequence ecosystem lives:

Info

The major aim of this framework is to define one set of Arguments that will be used by all simulation components. This is primarily for consistency: some parameters may be shared between several components, so it is appealing to only have to define them once.

Another aim is persistence: using the tree sequence data structure to save intermediate files, as well as to persist the entire simulation configuration into the tree sequence metadata allows for high reproducibility and easier debugging.

Usage

The framework provides various classes to either inherit from or use directly:

• Arguments
  • based on python dataclasses, inherit from Arguments and define arguments for all simulation components in dataclass-style
• PreCapSim, Recapitation, AddNeutralMut
  • abstract base classes for python-based simulation components
  • need to define the functions _assert_required and _make_ts in custom child classes
• FwdSim
  • run SLiM simulation, given .slim script defined in Arguments
• Simulation
  • define full simulation, given a set of components

Examples

For usage examples, see e.g.:

• https://github.com/clwgg/spatial_extinction_sim

Classes provided by ts-ipc

• Arguments
• PreCapSim
• Recapitation
• AddNeutralMut
• FwdSim
• Simulation