PhyloNet

PhyloNet infers phylogenetic networks from multilocus sequence alignments to reconstruct reticulate evolutionary histories such as hybridization and incomplete lineage sorting.

Key Features:

• Two-Step Inference Method: Infers networks on smaller subproblems and merges them into a comprehensive network using a divide-and-conquer strategy to improve scalability and computational efficiency.
• Handling Multilocus Data: Manages sequence alignments from multiple, unlinked loci to enable analysis of multilocus, multi-genome datasets.
• Hitting Set Problem Formulation: Formulates a Hitting Set version of the inference problem to minimize the number of trinets required for network construction and solves it with a heuristic.
• Performance Evaluation: Demonstrated on simulated and biological datasets with reported improvements in running time and accuracy relative to existing methods.

Scientific Applications:

• Reticulate evolution inference: Reconstructs evolutionary histories involving hybridization and other reticulation events while accounting for incomplete lineage sorting.
• Large-scale phylogenomic studies: Enables scalable phylogenetic network inference across numerous genomes and multilocus datasets previously limited by computational constraints.

Methodology:

Infer subnetworks from subsets of taxa, merge subnetworks into a single phylogenetic network, and apply a heuristic solution to a Hitting Set formulation to minimize the number of trinets required; the approach uses a divide-and-conquer two-step inference strategy and operates on multilocus, unlinked sequence alignments.