Lagrange-NG: The next generation of Lagrange

Area of Science:

• Biogeography
• Phylogenetics
• Computational Biology

Background:

• The Dispersion-Eradication and Cladogenesis (DEC) model is crucial for biogeographical analyses but computationally intensive due to matrix exponentiation.
• The exponential increase in states with more regions limits the scale of biogeographical studies.
• Existing tools like Lagrange face performance bottlenecks, restricting analyses to fewer regions.

Purpose of the Study:

• To develop a computationally efficient version of the Lagrange tool for biogeographical analyses.
• To overcome the limitations imposed by the number of regions in DEC model computations.
• To enable larger-scale biogeographical and phylogenetic research.

Main Methods:

• Redesigned the Lagrange tool, naming it Lagrange-NG, incorporating a novel matrix exponential computation method using Krylov subspaces.
• Enabled multithreading for significant performance improvements.
• Introduced a new metric for assessing differences between range inferences in phylogenetic trees.

Main Results:

• Lagrange-NG achieves speedups of up to 49x with multithreading and 26x on a single thread compared to the original Lagrange.
• The tool can now analyze datasets with up to 12 regions in approximately 18 minutes.
• Lagrange-NG shows improved software quality, with a SoftWipe score increasing from 5.5 to 7.8.

Conclusions:

• Lagrange-NG offers a substantial advancement in computational efficiency for biogeographical analyses using the DEC model.
• The tool's speed and improved accuracy metrics facilitate more extensive and complex phylogenetic and biogeographical research.
• Lagrange-NG is freely available, promoting wider adoption and scientific progress in the field.