Running _W ILD: the case for exploring mixed parameter sets in sensitivity analysis.

Area of Science:

• Phylogenetics and Evolutionary Biology
• Bioinformatics and Computational Biology

Background:

• Phylogenetic analyses aim for robust clades, often assessed using sensitivity analyses with the incongruence length difference (ILD) metric.
• Current practice typically involves selecting a single optimal parameter set to minimize ILD across all datasets.
• Divergent molecular evolution between ribosomal and protein-encoding genes (e.g., transversion bias in coding genes) suggests limitations of a single parameter set.

Discussion:

• This study explored the utility of mixed parameter sets in phylogenetic sensitivity analyses across five published datasets.
• Mixed parameter sets were compared against single optimal parameter sets, evaluating clade stability and congruence.
• The findings highlight how differing evolutionary dynamics of gene types necessitate varied analytical approaches.

Key Insights:

• In four out of five datasets, mixed parameter sets demonstrated superior performance over the single optimal parameter set.
• A single optimal parameter set remained superior only in cases with significant topological conflict between data partitions.
• Mixed parameter sets show promise for combining gene types with distinct evolutionary characteristics, like ribosomal and protein-encoding genes.

Outlook:

• Further investigation into mixed parameter sets could refine phylogenetic methods, especially for complex genomic datasets.
• This approach may enhance the accuracy of phylogenetic reconstructions when integrating data from ribosomal RNA and protein-coding genes.
• The study advocates for a more nuanced approach to parameter selection in phylogenetic analyses to improve clade robustness.