FuzzyClust

FuzzyClust performs fuzzy c-means clustering with automated parameter optimization to identify robust cluster structures in high-dimensional datasets such as DNA microarray and quantitative proteomics.

Key Features:

• Automatic Parameter Calculation: Automates calculation of key algorithm parameters, including the fuzzifier and the number of clusters.
• Optimal Fuzzifier Estimation: Estimates the optimal fuzzifier via a functional relationship based on dataset dimensionality and object count to avoid influence from random fluctuations and maximize resolution at the edge of randomness.
• Validation Indices for Cluster Number: Determines the optimal number of clusters using validation indices with a focus on minimum distance between centroids and computational efficiency relative to more resource-intensive methods.
• Randomized-dataset Evaluation: Evaluates clustering outcomes on randomized datasets to distinguish random noise from genuine cluster structure.

Scientific Applications:

• DNA microarray analysis: Identifies and characterizes cluster structures in DNA microarray expression datasets.
• Quantitative proteomics analysis: Detects and refines protein expression clusters in quantitative proteomics datasets.

Methodology:

Estimates the fuzzifier from a functional relationship of dataset dimensionality and object count, applies validation indices focusing on centroid distances to choose cluster number, and evaluates clustering outcomes on randomized datasets to select parameter values that prevent random noise being mistaken for meaningful structure while preserving genuine cluster resolution.