PhylteR

PhylteR identifies outlier sequences in phylogenomic datasets to improve the accuracy and concordance of gene trees and species tree inference.

Key Features:

• Automated Outlier Detection: Identifies sequences from individual gene trees that deviate from the general trend to flag potential errors.
• DISTATIS Framework: Implements DISTATIS, an extension of multidimensional scaling to three dimensions, to simultaneously compare multiple distance matrices derived from individual gene phylogenies representing evolutionary distances between species.
• Sensitivity and Precision: Demonstrates superior sensitivity and precision on simulated datasets for identifying outliers relative to existing methods.
• Robustness to ILS-Induced Incongruences: Shows insensitivity to incongruences induced by incomplete lineage sorting (ILS), reducing confounding by this biological process.
• Improvement of Concordance: Removal of identified outlier sequences improved concordance between gene trees and the species tree in a 14,463-gene, 53-species Carnivora dataset.
• Graphical Outputs: Generates graphical outputs to support visual characterization of datasets and identification of errors.

Scientific Applications:

• Species tree inference refinement: Improves species tree inference by detecting and enabling removal of anomalous gene sequences.
• Large-scale phylogenomic analyses: Applicable to high-throughput phylogenomic datasets where manual error detection is impractical.
• Signal-to-noise enhancement: Enhances the signal-to-noise ratio in phylogenomic datasets to increase reliability of evolutionary analyses.

Methodology:

Computes distance matrices from individual gene phylogenies and applies DISTATIS (a three-way multidimensional scaling extension) to simultaneously compare those matrices and identify sequences that deviate from the consensus of gene trees; performance was assessed on simulated datasets and on a 14,463-gene, 53-species Carnivora dataset.