Penalized likelihood phylogenetic inference: bridging the parsimony-likelihood gap

Area of Science:

• Phylogenetics
• Computational Biology
• Statistical Modeling

Background:

• Molecular data for phylogeny estimation is increasingly diverse and complex.
• Existing models and estimators struggle to handle this heterogeneity effectively.

Purpose of the Study:

• To propose a unified penalized likelihood (PL) framework for phylogeny estimation.
• To enable smooth control over the complexity of underlying phylogenetic models.
• To demonstrate the flexibility and broad applicability of the PL framework.

Main Methods:

• Development of a penalized likelihood (PL) framework.
• Demonstration using a four-taxon tree case.
• Investigation of the properties of the PL framework.

Main Results:

• The PL framework provides a unified approach encompassing various existing estimators (e.g., maximum parsimony, homogeneous likelihood, gamma mixture likelihood).
• Penalties allow for partitioning model complexity into controllable classes.
• Flexible control over model complexity is achieved.

Conclusions:

• The proposed PL framework offers a versatile and unified approach to phylogeny estimation.
• It allows for nuanced control over model complexity, accommodating diverse molecular data.
• This framework has the potential to improve the accuracy and robustness of phylogenetic analyses.