The demes Python package provides an API for defining, parsing, and sharing Demes demographic models. Applications can use demes to parse human-readable Demes YAML files into fully-resolved demographic models. In addition, demes provides convenient data structures to simplify manipulation of demographic models. If you find an error in the documentation or a bug in the software, please head to our git repository to report an issue or open a pull request.


Simulation is central to population genetics studies, and there are many great software packages out there for simulating sequencing data or computing expectations of diversity statistics under a wide range of demographic scenarios. This requires writing a formal description of the demographic model. Generally, each simulation software has its own syntax and style for defining the demography. Learning curves for new software can be steep and mistakes are easy to make, especially for complex demographic scenarios.

The Demes Specification aims to make defining demographic models more intuitive, less prone to error or ambiguity, and readily interchangeable between simulation platforms. Demographic models, which define populations (or demes), their properties, and relationships between them, are by convention written as a YAML file.


The following YAML file implements a two-epoch demographic history for a single deme, where the deme doubles in size 100 generations ago. See the Demes tutorial for a detailed introduction to writing Demes YAML files.

description: Single-population two-epoch demography.
time_units: generations
- name: deme0
  description: A deme that doubles in size 100 generations ago.
    - start_size: 1000
      end_time: 100
    - start_size: 2000
      end_time: 0

The YAML file can be loaded using demes, and then visually inspected using the demesdraw Python package.

import demes
import demesdraw

graph = demes.load("../examples/two_epoch.yaml")
<AxesSubplot:ylabel='time ago (generations)'>