AngClust: Angle Feature-Based Clustering for Short Time Series Gene Expression Profiles.

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Clustering gene expression data aims to group genes with similar expression patterns.
• Traditional methods often struggle with short-term gene expression data and identifying functionally relevant clusters.
• Existing algorithms may not effectively capture subtle changes in expression trends crucial for biological insights.

Purpose of the Study:

• To develop a novel clustering algorithm, AngClust, specifically designed for short-term gene expression data.
• To improve the identification of functionally significant gene clusters.
• To evaluate the performance of AngClust against established clustering methods.

Main Methods:

• Proposed AngClust, a clustering process utilizing angular features to represent gene expression trend changes between time points.
• Angular features at multiple time points are used to measure the similarity of expression trends between genes.
• Cluster significance was evaluated based on gene count, average correlation, fluctuation, and Gene Ontology (GO) term enrichment.

Main Results:

• AngClust demonstrated superior performance compared to Euclidean distance (ED) and dynamic time warping of correlation (DTW) on yeast gene expression data.
• AngClust achieved higher or equal ratios of GO and pathway term-enriched clusters compared to STEM and TMixClust.
• The method effectively captures similarities in expression trends, leading to more biologically meaningful clusters across human, mouse, and yeast datasets.

Conclusions:

• AngClust offers a novel and effective approach for clustering short-term gene expression data.
• The angular feature-based method enhances the identification of functionally relevant gene clusters.
• AngClust provides a valuable tool for analyzing time-series gene expression data in various organisms.