Current Protocols Vignette

This vignette is based on the published Current Protocols in Bioinformatics paper, Population genetic inference with Migrate. It is intended as the main self-contained tutorial for new users who want to understand how to start with MIGRATE, how to move to more realistic analyses, and how to use model selection once the basic workflow is clear.

Introductory use

The introductory stage should answer a narrow question with the smallest possible analysis:

• start with a small number of populations and loci
• use a simple migration model and a short exploratory run
• inspect whether the input file, parameter file, and output summaries behave as expected

At this stage, the goal is not to obtain the final scientific answer. The goal is to learn the mechanics of the program:

• how the data are formatted
• how a parameter file controls the run
• where to find estimates for population size and migration
• how to detect obvious setup problems before starting a larger analysis

For a first pass, users should keep the model simple, confirm that the chain mixes reasonably, and compare repeated short runs before increasing run length or model complexity.

More complicated setup

Once the basic workflow is stable, the next step is to move toward analyses that better reflect the biological question:

• increase the number of populations or loci
• choose mutation models and priors that fit the data type
• tune chain length, heating, and sampling settings for a harder inference problem
• decide whether the analysis is estimating migration only, divergence only, or both in combination

This is where MIGRATE becomes a research tool rather than a demonstration program. A more complicated setup should be built incrementally:

1. begin from a working simple configuration
2. change one major modeling choice at a time
3. rerun and inspect convergence and stability
4. only then move to the next increase in complexity

That sequence reduces the risk of confusing a biological signal with a configuration failure.

Model selection

Model selection is the final stage of the vignette. After users can run stable analyses, they can compare alternative demographic scenarios and ask which model is better supported by the data.

Typical comparisons include:

• restricted versus asymmetric migration
• panmixia versus structured populations
• alternative population-connection topologies
• simpler versus richer historical scenarios

The practical workflow is:

1. define a small set of biologically meaningful candidate models
2. fit each model with comparable run settings
3. compare marginal likelihood or related model-selection summaries reported by MIGRATE
4. interpret the preferred model together with parameter uncertainty, not in isolation

Users should avoid comparing too many poorly motivated models at once. A short, explicit set of alternatives usually gives a more interpretable result than a large fishing expedition.

Recommended progression

For most new users, the best path through the tutorials is:

1. read this vignette first
2. use the 2009 bird-population tutorial for historical context if needed
3. use the two-population tutorial for a focused model-comparison exercise
4. use the Zika tutorial for a larger multi-scenario example

This ordering gives a gradual path from basic execution to realistic model selection.