Extending bicluster analysis to annotate unclassified ORFs and predict novel functional modules using expression...

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Microarrays enable high-throughput gene expression analysis.
• Bicluster analysis identifies gene expression correlations in subsets of samples.
• Functional annotation of unclassified open reading frames (ORFs) remains a challenge.

Purpose of the Study:

• To extend bicluster analysis into a semi-supervised ORF annotation approach named BALBOA.
• To evaluate BALBOA's efficacy in classifying and annotating unclassified ORFs.
• To explore the potential of semi-supervised methods in gene expression analysis.

Main Methods:

• Developed BALBOA, a semi-supervised method integrating bicluster analysis with available annotations.
• Assessed BALBOA's performance using cross-validation against k-Nearest Neighbour (kNN) on three gene expression datasets.
• Applied BALBOA to assign functional annotations to unclassified yeast ORFs and predicted novel functional modules.

Main Results:

• BALBOA demonstrated efficacy in classifying unclassified ORFs across independent datasets.
• Functional predictions for yeast ORFs were evaluated using existing experimental and protein sequence data.
• The study successfully predicted the presence of novel functional modules within yeast.

Conclusions:

• Unsupervised methods like bicluster analysis can be extended to semi-supervised approaches for gene expression analysis.
• Semi-supervised methods show potential to outperform supervised approaches in ORF annotation.
• BALBOA offers insights into the functions of unclassified ORFs and their co-regulation patterns.