hstrat wordmark

hstrat wordmark

PyPi codecov Codacy Badge CI Read The Docs GitHub stars Zenodo JOSS

hstrat enables phylogenetic inference on distributed digital evolution populations


python3 -m pip install hstrat


hstrat serves to enable robust, efficient extraction of evolutionary history from evolutionary simulations where centralized, direct phylogenetic tracking is not feasible. Namely, in large-scale, decentralized parallel/distributed evolutionary simulations, where agents’ evolutionary lineages migrate among many cooperating processors over the course of simulation.

hstrat can

  • accurately estimate time since MRCA among two or several digital agents, even for uneven branch lengths

  • reconstruct phylogenetic trees for entire populations of evolving digital agents

  • serialize genome annotations to/from text and binary formats

  • provide low-footprint genome annotations (e.g., reasonably as low as 64 bits each)

  • be directly configured to satisfy memory use limits and/or inference accuracy requirements

hstrat operates just as well in single-processor simulation, but direct phylogenetic tracking using a tool likephylotrackpyshould usually be preferred in such cases due to its capability for perfect record-keeping given centralized global simulation observability.

Example Usage

This code briefly demonstrates,

  1. initialization of a population of HereditaryStratigraphicColumn of objects,

  2. generation-to-generation transmission of HereditaryStratigraphicColumn objects with simple synchronous turnover, and then

  3. reconstruction of phylogenetic history from the final population of HereditaryStratigraphicColumn objects.

from random import choice as rchoice
import alifedata_phyloinformatics_convert as apc
from hstrat import hstrat; print(f"{hstrat.__version__=}")  # when last ran?
from hstrat._auxiliary_lib import seed_random; seed_random(1)  # reproducibility

# initialize a small population of hstrat instrumentation
# (in full simulations, each column would be attached to an individual genome)
population = [hstrat.HereditaryStratigraphicColumn() for __ in range(5)]

# evolve population for 40 generations under drift
for _generation in range(40):
    population = [rchoice(population).CloneDescendant() for __ in population]

# reconstruct estimate of phylogenetic history
alifestd_df = hstrat.build_tree(population, version_pin=hstrat.__version__)
tree_ascii = apc.RosettaTree(alifestd_df).as_dendropy.as_ascii_plot(width=20)
              /--- 1
       /--+   \--- 3
       |  |
   /---+  \------- 2
   |   |
+--+   \---------- 0
   \-------------- 4

In actual usage, each hstrat column would be bundled with underlying genetic material of interest in the simulation — entire genomes or, in systems with sexual recombination, individual genes. The hstrat columns are designed to operate as a neutral genetic annotation, enhancing observability of the simulation but not affecting its outcome.

How it Works

In order to enable phylogenetic inference over fully-distributed evolutionary simulation, hereditary stratigraphy adopts a paradigm akin to phylogenetic work in natural history/biology. In these fields, phylogenetic history is inferred through comparisons among genetic material of extant organisms, with — in broad terms — phylogenetic relatedness established through the extent of genetic similarity between organisms. Phylogenetic tracking through hstrat, similarly, is achieved through analysis of similarity/dissimilarity among genetic material sampled over populations of interest.

Rather than random mutation as with natural genetic material, however, genetic material used by hstrat is structured through hereditary stratigraphy. This methodology, described fully in our documentation, provides strong guarantees on phylogenetic inferential power, minimizes memory footprint, and allows efficient reconstruction procedures.

See here for more detail on underlying hereditary stratigraphy methodology.

Getting Started

Refer to our documentation for a quickstart guide and an annotated end-to-end usage example.

The examples/ folder provides extensive usage examples, including

  • incorporation of hstrat annotations into a custom genome class,

  • automatic stratum retention policy parameterization,

  • pairwise and population-level phylogenetic inference, and

  • phylogenetic tree reconstruction.

Interested users can find an explanation of how hereditary stratigraphy methodology implemented by hstrat works “under the hood,” information on project-specific hstrat configuration, and full API listing for the hstrat package in the documentation.


If hstrat software or hereditary stratigraphy methodology contributes to a scholarly work, please cite it according to references provided here. We would love to list your project using hstrat in our documentation, see more here.


This package was created with Cookiecutter and the audreyr/cookiecutter-pypackage project template.