Crimp

Crimp aligns clusters across multiple clusterings using membership-coefficient (Q) matrices to mitigate label-switching in population structure analysis of multilocus genotype data and to enable comparison and averaging of replicate clusterings.

Key Features:

• Alignment of Clusters: Crimp aligns clusters based on matrices of membership coefficients (Q-matrices), commonly used to post-process outputs from Structure (Pritchard et al., 2000).
• Label-Switching Mitigation: It mitigates the label-switching problem in population structure analysis by aligning clusters across replicate clusterings consisting of the same number of clusters using a heuristic approach.
• Scalability and Efficiency: The software provides a fast, scalable heuristic for large datasets and an exact algorithm for smaller problem sizes to balance runtime and solution quality.
• Row-Specific Weights: Supports row-specific weights to allow customized alignment based on specific data characteristics.
• Compatibility with CLUMPP: Handles input and output files similar to CLUMPP (Jakobsson & Rosenberg, 2007) and evaluates solutions using both CLUMPP's objective functions and its own metrics.
• Benchmark Performance: Benchmark analyses demonstrate superior runtime and quality measures compared to alternative tools, particularly for larger datasets.

Scientific Applications:

• Population Structure Analysis: Facilitates assignment of individuals to a prespecified number of ancestral populations using model-based clustering and comparison of replicate analyses.
• Post-processing of Structure Outputs: Enables comparison and generation of averaged clusterings from outputs of Structure and similar clustering programs using Q-matrices.

Methodology:

Aligns clusters using membership-coefficient (Q) matrices; employs a heuristic approach for large datasets and an exact algorithm for smaller ones; supports row-specific weights; evaluates solutions with CLUMPP objective functions and tool-specific metrics.