A fast likelihood solution to the genetic clustering problem.

Area of Science:

• Population genetics
• Molecular ecology
• Conservation biology

Background:

• Identifying genetic clusters is crucial for understanding population structure and evolutionary processes.
• Existing methods for genetic clustering are often computationally intensive or lack interpretability.

Purpose of the Study:

• To introduce snapclust, a novel, fast, and accurate method for genetic clustering and hybrid detection.
• To provide a computationally efficient tool for analyzing genetic diversity in natural populations.

Main Methods:

• snapclust employs a maximum-likelihood approach using the Expectation-Maximisation (EM) algorithm.
• Combines geometric approaches with fast likelihood optimization for efficiency.
• Includes goodness-of-fit statistics to guide cluster number selection.

Main Results:

• snapclust demonstrates comparable performance to gold-standard methods for clustering and hybrid detection.
• Shows advantages in identifying complex hybrids after multiple backcrosses.
• Is orders of magnitude faster than existing model-based methods.

Conclusions:

• snapclust offers a flexible and computationally efficient solution for genetic clustering and hybrid detection.
• Easily integrated into existing R-based genetic analysis pipelines.
• A valuable addition to the toolbox for studying genetic diversity in natural populations.