gstudio

An R package for the spatial analysis of population genetic data, including Population Graph construction, simulation, and directional gene flow analysis.

For a complete overview of the package, see the full documentation at http://dyerlab.github.io/gstudio/

To install from GitHub, fire up R and run:

remotes::install_github("dyerlab/popgraph")
remotes::install_github("dyerlab/gstudio")

Key functions

Population simulation

• make_populations() / migrate() / mixed_mating() — forward-time individual-based simulation
• partition() / Fst() — summary statistics

Population Graph construction

• popgraph(x, groups) — constructs a Markov Random Field graph from multilocus genetic data; returns an undirected weighted popgraph / igraph object

Directional gene flow (genetic gravity)

• graph_asymmetries(graph) — takes any weighted undirected popgraph and adds three new attributes without altering the topology:
  • V(graph)$bandwidth — per-node local bandwidth $b_i$ = mean edge weight (strength / degree)
  • E(graph)$w_away, E(graph)$w_to — directional Gaussian kernel weights for each edge
  • E(graph)$delta — asymmetry index $\Delta_{ij} = w_{i \to j} - w_{j \to i}$; positive values indicate net gene flow from the first to the second endpoint

The bandwidth estimator is grounded in the Markov structure of the Population Graph: conditional independence across non-adjacent nodes means the neighborhood edge weights are sufficient statistics for the decay of genetic signal at each node, so no global distance matrix or iterative calibration is required.

Contributing

If you would like to contribute to this package or have any questions regarding its construction, use, or additional functionality, please contact Rodney J. Dyer or visit his lab page at http://dyerlab.org