COBRAC

COBRAC performs convex biclustering by reformulating biclustering as a convex optimization problem and applying iterative compression to identify simultaneous groups of observations and features in gene expression and genomic datasets.

Key Features:

• Convex biclustering formulation: Reformulates biclustering as a convex optimization problem to improve robustness and solution quality.
• Iterative compression of the data matrix: Compresses the data matrix iteratively to reduce problem size and computational cost.
• Compression along the solution path: Applies iterative compression to the solution space/path to accelerate convergence and computation.
• Computational efficiency: Reduces computing time compared with conventional biclustering algorithms through compression-based acceleration.
• Scalability to large gene expression and genomic datasets: Enables processing of extensive high-throughput biological data.

Scientific Applications:

• Gene expression analysis: Identifies biclusters in gene expression datasets to reveal subsets of genes and samples with shared patterns.
• Gene regulation and co-expression network studies: Detects biologically relevant patterns useful for studying gene regulation and co-expression networks.
• Analysis of complex biological systems: Extracts simultaneous grouping patterns across observations and features in large-scale genomic data.

Methodology:

Reformulates the biclustering problem as a convex optimization task and solves it using an iterative compression technique that reduces both the data matrix and the solution space along the solution path.