Bi-Force: large-scale bicluster editing and its application to gene expression data biclustering

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• The rapid growth of biological data necessitates advanced bioinformatics approaches for analysis.
• Biclustering is a key technique for identifying local patterns in high-dimensional biological datasets, such as gene expression data.

Purpose of the Study:

• To introduce Bi-Force, a novel heuristic for biclustering based on a weighted bicluster editing model.
• To evaluate the performance of Bi-Force against existing biclustering algorithms using established protocols and diverse datasets.

Main Methods:

• Developed Bi-Force using a weighted bicluster editing model for arbitrary biological entities and pairwise similarities.
• Compared Bi-Force with two algorithms in the BiCluE package and eight external tools (FABIA, QUBIC, etc.).
• Utilized synthetic and large-scale gene expression datasets (Gene Expression Omnibus) for evaluation, followed by Gene Ontology enrichment analysis.

Main Results:

• Bi-Force demonstrated superior performance compared to existing tools when evaluated using the Eren et al. protocol.
• The bicluster editing foundation of Bi-Force proved more powerful than strict biclustering methods.
• Gene Ontology enrichment analysis confirmed the biological relevance of biclusters identified by Bi-Force.

Conclusions:

• Bi-Force offers a more powerful approach to biclustering due to its distinct theoretical foundation in bicluster editing.
• The method effectively identifies biologically relevant patterns in gene expression data.
• Bi-Force is publicly available as part of the open-source BiCluE software package.