Phylogenetic Branch Length Data from Six Problematic Datasets

Data from a study by Chi Zhang (2019) investigates the robustness of compound Dirichlet priors for Bayesian branch length inference. The dataset comprises six problematic phylogenetic datasets originally analyzed by Brown et al. (2010). It was generated by modifying MrBayes 3.1.2 to incorporate the new priors and comparing posterior estimates against maximum likelihood results.

Use Cases

• Compare posterior tree length estimates from compound Dirichlet priors against maximum likelihood estimates for the six datasets.
• Analyze the robustness of posterior tree lengths to changes in the parameter values of the compound Dirichlet priors.
• Investigate branch-length overestimation by comparing results from default exponential priors and the proposed compound Dirichlet priors.
• Replicate the study's analysis of properties of Bayesian branch-length estimation using the six problematic datasets.

Strengths

• Dataset is derived from six specific, previously identified problematic phylogenetic datasets for branch-length estimation.
• Results demonstrate the compound Dirichlet priors produced posterior estimates much closer to maximum likelihood estimates than default priors.
• Posterior tree lengths were robust to changes in prior parameter values over several orders of magnitude.

Limitations

• The dataset is limited to six specific phylogenetic datasets, which may not represent all empirical data scenarios.
• The exact data format, size, and specific variables (e.g., individual branch lengths, tree topologies) are not detailed in the input.
• The data is from a 2019 study and may not reflect the most recent methodological developments in Bayesian phylogenetics.

Provenance

Source

Chi Zhang, via Dryad.

Collection Method

Generated by modifying MrBayes 3.1.2 to incorporate compound Dirichlet priors and re-analyzing six existing phylogenetic datasets.

Freshness

Last updated 2019-06-26.