Vector Encoding of Phylogenetic Trees by Ordered Leaf Attachment.

Area of Science:

• Computational Biology
• Machine Learning
• Phylogenetics

Background:

• Recent interest in vector encodings for phylogenetic trees to bridge phylogenetics and machine learning.
• Existing methods may lack efficiency or unique properties for tree representation.

Purpose of the Study:

• To present a novel, simple, and unique method for encoding binary, rooted phylogenetic tree topologies as integer vectors.
• To establish a new tree distance metric based on vector encoding and compare it with existing metrics.

Main Methods:

• Developed the "ordered leaf attachment" (OLA) encoding method.
• Demonstrated linear time complexity for OLA encoding and decoding relative to the number of leaf nodes.
• Analyzed the properties of the OLA encoding and the induced tree distance.

Main Results:

• The OLA method provides a unique integer vector encoding for phylogenetic trees.
• Encoding and decoding processes exhibit linear time complexity.
• The OLA encoding defines a simply-described subset of integer sequences corresponding to valid tree topologies.
• Investigated the relationship between the OLA-induced distance and established tree distances like NNI and SPR.

Conclusions:

• The OLA method offers an efficient and unique approach to vector encoding of phylogenetic trees.
• This method facilitates the application of machine learning techniques to phylogenetic analysis.
• The OLA-induced distance provides a new perspective for comparing phylogenetic trees.