ValWorkBench

ValWorkBench provides cluster-number prediction and validation for biological datasets, especially microarray data, by implementing eleven external and internal validation measures and heuristic approximations.

Key Features:

• Extensive Validation Measures: Implements eleven external and internal metrics including Adjusted Rand Index, Figure of Merit, Gap Statistics, Within Cluster Sum Square, and Consensus Clustering.
• Heuristic Approximations: Provides heuristic approximations for some validation measures to reduce computational cost.
• Extensible Architecture: Offers a library architecture that allows integration of additional validation measures.
• Software Abstraction: Employs modular, abstracted components to enable reuse of validation implementations.
• Clustering Compatibility: Applies validation metrics to outputs from diverse clustering algorithms.

Scientific Applications:

• Microarray clustering analysis: Assess and predict the optimal number of clusters in microarray datasets.
• Cluster solution assessment: Compare and evaluate clustering solutions using external and internal metrics and consensus clustering.

Methodology:

Computes external and internal cluster validation metrics and heuristic approximations that are applied to outputs of various clustering algorithms.