matrix-clustering

matrix-clustering clusters position-specific scoring matrices (PSSMs) to group similar transcription factor binding motifs (TFBMs) and reduce redundancy in motif databases.

Key Features:

• Hierarchical Clustering: Applies hierarchical clustering to sets of PSSMs to identify groups of similar motifs and generate cluster trees.
• Branch Motif Construction: Constructs branch motifs at internal nodes by merging the matrices from all descendant nodes.
• Tree Generation: Produces phylogenetic-like trees for each cluster to represent relationships among motifs.
• Motif Alignment Visualization: Aligns TFBMs to visualize structural similarities and differences among clustered motifs.
• Multiple Collection Handling: Processes multiple motif collections simultaneously, including motifs from diverse databases or ChIP-seq results.
• Redundancy Reduction: Aggregates similar motifs into non-redundant collections to simplify interpretation of motif discovery outputs.

Scientific Applications:

• Genomics and Transcriptional Regulation: Analysis of transcription factor binding motifs to investigate regulatory patterns across genomes.
• Integration of Motif Discovery Results: Simplifies interpretation of combined outputs from multiple motif discovery tools by grouping biologically related motif variants.
• Large-scale Motif Database Curation: Clusters extensive motif sets (e.g., 24 databases containing over 7,500 motifs) to group motifs belonging to the same transcription factor families.

Methodology:

Accepts one or several sets of PSSMs as input; utilizes hierarchical clustering to identify motif clusters; produces trees per cluster and constructs branch motifs at each node by merging descendant matrices; aligns TFBMs for visualization.