selectKSigs

selectKSigs suggests the number of mutational signatures in a collection of somatic mutations by calculating the cross-validated perplexity score.

Key Features:

• Mutational signature number selection: The package provides a data-driven approach to determine the optimal number of mutational signatures through statistical evaluation.
• Cross-validation: The method employs cross-validation to assess model performance and prevent overfitting when estimating signature numbers.
• Perplexity-based measure: The package uses perplexity as a quantitative metric to evaluate and compare different signature number configurations.

Scientific Applications:

• Somatic mutation analysis: The package enables researchers to determine the appropriate complexity of mutational signature models for cancer genomics studies.
• Mutational signature deconvolution: selectKSigs supports downstream signature extraction workflows by providing an informed choice for the number of signatures to extract.

Methodology:

The method calculates cross-validated perplexity scores to evaluate and suggest the number of mutational signatures present in somatic mutation data.