Modeling and analysis of gene expression time-series based on co-expression

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Gene expression time-series analysis is crucial for understanding dynamic biological processes.
• Existing clustering methods may not fully capture temporal patterns or biological relevance.
• Smooth characterization of time-series data is needed for accurate modeling.

Purpose of the Study:

• To introduce a novel approach for modeling and clustering gene expression time-series.
• To develop a new co-expression coefficient for evaluating temporal similarity.
• To improve the biological relevance of gene expression data clustering.

Main Methods:

• Utilized radial basis function neural networks for generalized time-series characterization.
• Defined a novel co-expression coefficient based on temporal shapes and time point distribution.
• Employed a fuzzy clustering algorithm with the proposed co-expression metric for grouping profiles.

Main Results:

• Demonstrated the effectiveness of the proposed metric and method on artificial and real gene expression data.
• Showcased improved grouping of temporal profiles compared to standard methods.
• The novel co-expression coefficient outperformed the traditional correlation coefficient in producing biologically relevant clusters.

Conclusions:

• The proposed method provides a robust framework for modeling and clustering gene expression time-series.
• The novel co-expression coefficient enhances the biological interpretability of clustering results.
• This approach offers a significant advancement in analyzing dynamic gene expression patterns.