Bicluster

Bicluster identifies coherent gene clusters within microarray gene expression data by finding subsets of genes that exhibit similar patterns across specific experimental conditions.

Key Features:

• Seed-Based Algorithm: Bicluster employs a seed-based algorithm that exhaustively selects combinations of genes and conditions as initial seeds to generate candidates.
• Candidate Bicluster Tables: Each candidate table comprises two columns: a gene set and the conditions under which that gene set exhibits dissimilar expression relative to the seed.
• Exhaustive Testing: The algorithm exhaustively tests combinations of genes and conditions to ensure comprehensive coverage of possible bicluster configurations.
• Gene Filtering: Genes are filtered based on the maximum number of dissimilar conditions they exhibit to refine the candidate table.
• Condition Grouping: Rows sharing identical dissimilar conditions are grouped together to streamline identification of coherent gene sets.
• Sorting and Threshold Testing: The table is sorted in ascending order by the number of dissimilar conditions and tested sequentially against a minimum bicluster formation threshold.
• Iterative Output Removal: Identified biclusters are output and the corresponding rows are removed iteratively until no further biclusters remain.

Scientific Applications:

• Condition-Specific Gene Cluster Discovery: Identifies gene subsets that show coherent expression patterns under specific experimental conditions in microarray datasets.
• Elucidation of Gene Regulatory Mechanisms: Enables analysis of condition-dependent gene networks and regulatory interactions through additive biclusters.
• Functional Interpretation of Gene Sets: Reveals how different gene sets contribute to biological functions by exposing condition-specific expression coherence.

Methodology:

Generate candidate bicluster tables from seeds, exhaustively test combinations of genes and conditions, filter genes by maximum dissimilar conditions, group rows with identical dissimilar conditions, sort rows ascending by number of dissimilar conditions, test rows sequentially against a minimum threshold, and iteratively output biclusters while removing corresponding rows.