Ensembl query-based biclustering

Ensembl query-based biclustering identifies non-redundant biclusters in gene expression compendia by performing ensemble query-based biclustering across multiple query genes and parameter settings and consolidating results via a consensus matrix.

Key Features:

• Query-Based Biclustering: Accepts gene lists or individual genes as queries and finds genes with expression profiles similar to the average profile of the query list.
• Handling Non-Coexpressed Genes: Processes each gene individually as a separate query to address long gene lists that lack mutual coexpression.
• Reduction of Redundancy: Integrates results from multiple parameter settings to reduce redundant biclustering outcomes produced by varying bicluster sizes.
• Consensus Matrix Design: Constructs a consensus matrix that statistically merges biclustering results across different query-genes and parameter settings.
• Non-Redundant Bicluster Identification: Clusters the consensus matrix to extract distinct, non-redundant biclusters that maximize informational content from query-based results.

Scientific Applications:

• Microbial genomics: Enables identification of expression-based biclusters relevant to microbial gene expression compendia.
• Cancer research: Supports analysis of extensive gene lists from cancer studies to reveal coherent expression modules.
• Escherichia coli case study: Has been applied to an Escherichia coli compendium to demonstrate identification of biologically relevant biclusters.

Methodology:

Accepts gene lists as queries; performs query-based biclustering for each gene and across multiple parameter settings; merges results into a consensus matrix using statistical methods; clusters the consensus matrix to identify non-redundant biclusters.