A New Phylogenetic Inference Based on Genetic Attribute Reduction for Morphological Data.

Area of Science:

• Phylogenetics
• Computational Biology
• Data Science

Background:

• Phylogenetic tree reconstruction is crucial for understanding evolutionary relationships.
• Missing data in morphological datasets poses a significant challenge, leading to instability in phylogenetic inference.

Purpose of the Study:

• To develop a novel phylogenetic inference method robust to missing data in morphological datasets.
• To enhance the accuracy and stability of phylogenetic trees.

Main Methods:

• A concept decision tree (CDT) method combined with attribute reduction was developed.
• An initial phylogenetic seed tree was constructed using species with complete data.
• A top-down approach with concept-sample templates and a novel matching algorithm was employed for tree construction.

Main Results:

• The CDT method demonstrated high accuracy (86.5%) even with increasing proportions of missing data.
• The CDT method outperformed maximum parsimony, maximum likelihood, and Bayesian inference in terms of accuracy and stability.
• The developed algorithm successfully established reliable phylogenetic trees.

Conclusions:

• The concept decision tree (CDT) method offers a stable and accurate approach for phylogenetic inference from morphological data with missing values.
• This method provides a significant improvement over existing techniques when dealing with incomplete datasets.