Clumppling

Clumppling aligns replicate mixed-membership unsupervised clustering solutions in population genetics by formulating cluster alignment as an integer linear programming problem to obtain optimal alignments across replicates.

Key Features:

• Optimal Alignment: Uses integer linear programming (ILP) to find globally optimal alignments of clustering solutions across multiple replicates.
• Computation Efficiency: Produces superior alignment results with reduced computation time compared with existing methods such as Pong and Clumpak.
• Flexibility in Cluster Numbers (K): Supports alignment of replicates that have differing arbitrary values for the number of clusters (K).

Scientific Applications:

• Population Genetics: Facilitates interpretation of mixed-membership models to study genetic structure and variation within populations.
• Replicate Integration: Enables combining and comparing clustering solutions from multiple replicates to improve inference of genetic diversity and evolutionary patterns.

Methodology:

Clumppling frames the cluster alignment problem as an integer linear programming task that maximizes a predefined objective function and embeds the problem within combinatorial optimization frameworks.