[Paper Review] An empirical guide for producing a dated phylogeny with treePL in a maximum likelihood framework

Kévin J. L. Maurin|arXiv (Cornell University)|Aug 17, 2020

Genomics and Phylogenetic Studies11 references35 citations

TL;DR

This paper provides a step-by-step protocol for obtaining a dated phylogeny with treePL within a maximum likelihood framework, includes reproducible data, and compares treePL results to BEAST2-built phylogenies.

ABSTRACT

treePL uses a penalised likelihood approach to produce a dated phylogeny in a maximum likelihood framework. Since its publication in 2012, few resources have been developed to explain how to use it properly. In this guide, I provide a step-by-step protocol for producing a dated phylogeny using treePL, based on my experience building a large dated phylogeny with it and conducting additional tests on a smaller phylogeny. I also provide the necessary data to reproduce one of the example phylogenies presented. I compare these treePL phylogenies to BEAST2-built counterparts. Even though I cannot explain precisely how treePL works, the evidence discussed in this guide suggest that the empirical protocol presented is reliable.

Motivation & Objective

• Provide a practical protocol for using treePL to generate dated phylogenies within a maximum likelihood framework.
• Share empirical experiences from building a large dated phylogeny and testing on a smaller phylogeny.
• Offer reproducible data to replicate example phylogenies and facilitate method evaluation.

Proposed method

• Describe a step-by-step protocol for treePL usage in ML dating.
• Present empirical tests and comparisons with BEAST2-derived phylogenies.
• Offer supplementary data to reproduce at least one example phylogeny.
• Discuss reliability of the empirical protocol based on observed results.

Experimental results

Research questions

• RQ1 Can treePL produce reliable dated phylogenies within a maximum likelihood framework when following an explicit empirical protocol?
• RQ2 How do treePL-derived dated phylogenies compare to BEAST2-built counterparts in this context?
• RQ3 What data and tests are needed to replicate the example phylogenies and assess robustness?

Key findings

• An empirical protocol with step-by-step guidance can yield dated phylogenies using treePL.
• TreePL phylogenies are compared to BEAST2 results, informing reliability and differences between methods.
• Supplementary data enables replication of the example phylogenies.