ClusterSignificance

ClusterSignificance assesses the statistical significance of cluster separations in high-throughput genomic and molecular biology datasets using one-dimensional projection and permutation testing to compute p-values.

Key Features:

• Cluster assessment: Evaluates whether observed separations between clusters differ significantly from expectations under random labeling.
• One-dimensional projection: Projects multi-dimensional data points onto a one-dimensional line to simplify separation measurement.
• Scoring and probability evaluation: Computes separation scores for projected clusters and derives probabilities, including p-values, for observed scores.
• Permutation testing: Uses permutation methods to generate a null distribution of cluster separations for statistical inference.
• Bioconductor integration: Implemented as an R package within the Bioconductor ecosystem to enable interoperability with other Bioconductor packages.

Scientific Applications:

• Genomics data analysis: Assess significance of clusterings in high-throughput genomic datasets such as gene expression profiles.
• Molecular biology studies: Evaluate cluster separations in molecular datasets to support interpretation of experimental groupings.
• Biological interpretation of groupings: Provide statistical evidence for cluster-based hypotheses related to genetic variation and disease mechanisms.

Methodology:

Project multi-dimensional data to one dimension, compute separation scores for clusters, and apply permutation testing to generate a null distribution and calculate p-values.