ML or NJ-MCL? A comparison between two robust phylogenetic methods.

Area of Science:

• Computational Biology
• Phylogenetics
• Bioinformatics

Background:

• Large-scale gene sequencing enables reconstruction of evolutionary histories (phylogenies) from extensive datasets.
• Accurate and computationally efficient phylogenetic reconstruction methods are crucial for analyzing these large datasets.

Purpose of the Study:

• To compare the accuracy of the Neighbor-Joining-Maximum-Composite-Likelihood (NJ-MCL) and Maximum Likelihood (ML) methods for phylogenetic reconstruction.
• To evaluate the performance of these methods using biologically realistic computer simulations.

Main Methods:

• Biologically realistic computer simulations were employed to assess phylogenetic inference accuracy.
• The concatenation approach, involving progressive gene addition, was utilized.
• Accuracy of NJ-MCL and ML methods were directly compared.

Main Results:

• NJ-MCL demonstrates computational efficiency and outperforms the Neighbor-Joining (NJ) method.
• However, the Maximum Likelihood (ML) method exhibits significantly higher accuracy compared to NJ-MCL.
• NJ-MCL is preferred for reconstructing very large phylogenies (thousands of taxa) due to its speed.

Conclusions:

• ML is recommended for phylogenetic analyses involving up to several hundred species due to its superior accuracy.
• For grand-scale phylogenies with thousands of taxa, the computationally efficient NJ-MCL method is a suitable alternative.