Inference Algorithms

ARGscape provides access to multiple spatial and temporal inference algorithms for estimating ancestral locations and node times. This guide explains each algorithm, when to use them, and their accuracy vs. speed tradeoffs.

Overview

Inference algorithms in ARGscape fall into two categories:

Category	Purpose	Algorithms
Spatial	Infer geographic locations of ancestors	FastGAIA, Geoancestry (GAIA), SPARG, Spacetrees, Midpoint
Temporal	Infer times (ages) of ancestral nodes	Tsdate

Note

Availability by interface:

• Python API (argscape.infer()): Midpoint, FastGAIA, GAIA, Tsdate

• CLI (argscape infer): Midpoint, FastGAIA, GAIA, SPARG

• Web App: All algorithms including SPARG and Spacetrees

Spatial Inference Algorithms

FastGAIA

Parameters

Parameter	Type	Default	Description
weight_span	bool	True	Weight by genomic span of edges
weight_branch_length	bool	True	Weight by branch length (time)

---

GAIA Quadratic / Linear

Parameters

Parameter	Type	Default	Description
use_branch_lengths	bool	False	Incorporate branch lengths in parsimony calculation

---

SPARG

Note

SPARG is available in the CLI and web app only. It is not exposed in the Python API.

Output Information

SPARG returns additional information:

• dispersal_rate: Estimated dispersal rate matrix

• num_inferred_locations: Number of ancestors located

---

Spacetrees

Note

Spacetrees is available in the web app only. It is not exposed in the Python API or CLI.

Parameters

Parameter	Type	Default	Description
time_cutoff	float	None	Only infer locations for ancestors younger than this
ancestor_times	list	None	Specific times to locate ancestors (None = all unique node times)
use_importance_sampling	bool	True	Weight trees by coalescent probability
require_common_ancestor	bool	True	Only use trees where all samples coalesce
Ne	float	None	Effective population size (constant)
Ne_epochs	list	None	Time boundaries for varying Ne
Nes	list	None	Ne values for each epoch
use_blup	bool	False	Use BLUP instead of MLE
blup_var	bool	False	Also return variance estimates

---

Midpoint Inference

Parameters

Parameter	Type	Default	Description
weight_by_span	bool	True	Weight by genomic span of edges
weight_branch_length	bool	False	Weight by branch length

---

Temporal Inference

Tsdate

Parameters

Parameter	Type	Default	Description
mutation_rate	float	1e-8	Per-site, per-generation mutation rate
preprocess	bool	True	Apply preprocessing steps
remove_telomeres	bool	False	Remove telomeric regions
minimum_gap	float	None	Minimum gap size for splitting
split_disjoint	bool	True	Split disjoint trees

---

See Also

• argscape.visualize() - Main visualization API

• 3D Spatial Visualization - 3D spatial visualization tutorial

• argscape infer - CLI for running inference