Skip to main content

Advanced tutorials

Change the controls beyond the defaults, and judge the result in Diagnostics before you interpret it.
  • You’ll learn What each advanced control changes, its default, when to change it, and what to check afterwards.
  • Time 20–40 minutes per family
  • Prerequisites The matching beginner tutorial: Phylo traits, Ancestral state reconstruction, Diversification or SSE models. Guane running, with the example loaded and matched in that module’s Data card.

Choose a family

Advanced phylo traits

Fixed or estimated covariance, ML versus REML, and optimizer search.

  • Keep correlation parameter fixed
  • Estimation method
  • Optimization starts
  • Maximum rate per branch-length unit

Advanced ancestral states

Root priors, fixed rate matrices, sampled histories and measurement error.

  • Root probabilities (pi)
  • Supply fixed Q
  • Stochastic mapping (fixed fitted rates)
  • Known standard-error column

Advanced diversification

Incomplete sampling, rate bounds, profile intervals and missing species.

  • Random extant sampling fraction
  • Rate upper bound (blank = 100 / crown age)
  • Calculate profile likelihoods
  • Missing extant species

Advanced SSE models

State sampling, tied rates, optimizer verification and QuaSSE numerics.

  • Sampling fraction for state 0
  • BiSSE parameter table (CSV)
  • Verify with an independent optimizer
  • Trait grid bins

How to read a control card

Each family page describes its controls in cards like this one:

Root probabilities (pi)

Sets the prior probability of each state at the root of a discrete reconstruction.

Where
Discrete traits Advanced analysis controls Root probabilities (pi)
Default
Equal weights
Change it when
You have independent evidence about the root state, for example from fossils or an outgroup.
Check after
Compare the root node’s probabilities with your weights. If they match exactly, the data added nothing at the root.
  • Where is the path to the control in the app.
  • Default is what Guane uses if you change nothing.
  • Change it when names the situation that justifies a change.
  • Check after tells you what to read, usually in Diagnostics, before you trust the new result.

Judge before you interpret

Five habits apply to every advanced control:

  1. Read Diagnostics first. Warnings sit there, not next to the estimates.
  2. Compare all optimizer starts. If starts reach different likelihoods, the best one may not be the global optimum.
  3. A rate at a bound is not an estimate. It tells you the data cannot pin that rate down.
  4. Missing weights are deliberate. Guane withholds weights and intervals when checks fail; do not rebuild them by hand.
  5. Rerun after every change. Changing data or settings clears the affected result.

See Uncertainty and reporting, Quantities that must not be conflated and Minimum information to report.

Your own data

Many advanced controls take pasted tables or matrices. Custom input formats gives the exact format for each box, and Troubleshooting pasted inputs covers rejected tables.