treelikeness

treelikeness assesses the extent to which a phylogenetic alignment conforms to the treelikeness assumption by quantifying the proportion of edge weights in a comprehensive phylogenetic network that are represented in a bifurcating phylogenetic tree derived from the same alignment.

Key Features:

• Tree proportion statistic: Quantifies the fraction of phylogenetic-network edge weights captured by a bifurcating tree derived from the same alignment.
• Parametric bootstrap test: Extends the tree proportion statistic into a formal statistical test of treelikeness using a parametric bootstrap approach.
• Detection of non-treelikeness sources: Targets violations of treelikeness arising from recombination, introgression, incomplete lineage sorting, and systematic errors.
• Comparative evaluation: Performance has been assessed via extensive simulations and compared against alternative methods.
• Complementary analysis: Intended for combined use with mosaic-detection methods such as 3Seq to provide more comprehensive assessment of alignments.

Scientific Applications:

• Testing treelikeness of alignments: Statistically evaluate whether a multiple-sequence alignment can be accurately described by a single bifurcating tree.
• Recombination and introgression detection: Identify alignments where recombination or introgression produce non-treelike signal.
• Incomplete lineage sorting and systematic error assessment: Flag alignments where incomplete lineage sorting or systematic biases violate treelikeness assumptions.
• Guiding phylogenetic inference: Inform decisions on whether tree-based phylogenetic methods are appropriate for a given data set.
• Method benchmarking: Use simulation-based comparisons to benchmark treelikeness assessment against other approaches.

Methodology:

Compute the tree proportion as the proportion of edge weights in a comprehensive phylogenetic network represented in a bifurcating tree derived from the same alignment, extend this statistic into a formal test using a parametric bootstrap, and evaluate performance via extensive simulations comparing to other methods.