A robust approach based on Weibull distribution for clustering gene expression data.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression data analysis commonly employs clustering techniques.
• Existing methods often focus on distance metrics, with fewer utilizing distribution similarities.
• Evaluating clustering based on functional consistency is crucial with growing biological annotations.

Purpose of the Study:

• To introduce a novel clustering method for gene expression data based on distribution similarities.
• To assess the performance of the proposed method against established algorithms.
• To demonstrate the method's capability in handling incomplete gene expression datasets.

Main Methods:

• Proposed the Weibull Distribution-based Clustering Method (WDCM).
• Modeled individual gene expressions as random variables following Weibull distributions.
• Clustered genes based on the similarity of their Weibull distribution parameters.
• Validated performance using functional annotation data (Gene Ontology) and Adjusted Rand Index.

Main Results:

• WDCM successfully clustered three cancer gene expression datasets (lung, B-cell lymphoma, bladder).
• WDCM demonstrated higher functional annotation ratios compared to k-means and Self-Organizing Map (SOM).
• Comparative analysis confirmed superior clustering performance of WDCM over k-means and SOM.

Conclusions:

• WDCM generates clusters with enhanced functional consistency.
• The method is robust and effective for gene expression data with missing values.
• WDCM offers a valuable alternative for gene expression data clustering, especially with incomplete datasets.