Integrative clustering by nonnegative matrix factorization can reveal coherent functional groups from gene profile...

Area of Science:

• Molecular Biology
• Bioinformatics
• Computational Biology

Background:

• Advancements in molecular biology generate vast, diverse datasets for gene profiling and function prediction.
• Integrating heterogeneous gene profile data into a unified prediction model remains a challenge.
• Profile-based clustering is a key technique for identifying groups of genes with similar expression patterns.

Purpose of the Study:

• To develop a robust method for fusing gene clusters derived from multiple, heterogeneous data sources.
• To improve the accuracy and quality of gene function prediction through integrated clustering.
• To demonstrate the efficacy of the proposed fusion technique using real-world biological data.

Main Methods:

• Developing individual gene clusters from each available data source.
• Employing nonnegative matrix factorization (NMF) to fuse these separate clusters.
• Utilizing gene profile data from the budding yeast Saccharomyces cerevisiae for validation.

Main Results:

• The proposed technique successfully integrates heterogeneous gene profile datasets.
• The fused clusters exhibit higher quality and reveal gene relationships not apparent from merged profiles.
• Demonstrated successful application in Saccharomyces cerevisiae gene profiling.

Conclusions:

• The NMF-based fusion of independently derived gene clusters is an effective strategy for integrating diverse biological data.
• This approach enhances the ability to infer gene function and identify biologically relevant gene groups.
• The method offers a significant improvement over traditional approaches that merge data prior to clustering.