Ensemble NMF

Ensemble NMF performs ensemble non-negative matrix factorization to cluster protein interaction networks and identify localized functional protein complexes.

Key Features:

• Aggregation of Matrix Factorizations: Integrates multiple non-negative matrix factorizations (NMF) to produce nuanced clustering outcomes.
• Soft Hierarchy Clustering: Generates a "soft" hierarchical clustering that represents flexible cluster memberships.
• Discovery of Localized Structures: Detects small, localized structures that correspond to known functional groupings of protein complexes.
• Applicability to Binary Interactions: Operates effectively on high-quality assemblies of binary protein interactions and proteome-wide datasets.

Scientific Applications:

• Protein Interaction Analysis: Clustering and analysis of protein-protein interaction networks using high-quality binary interaction assemblies and proteome-wide studies.
• Functional Grouping Identification: Identification of known functional groupings within yeast protein complexes.
• Putative Function Assignment: Assignment of putative functions to uncharacterized proteins, exemplified by the suggestion that YNR024W may be a component of the exosome.

Methodology:

Aggregates multiple non-negative matrix factorizations to produce a soft hierarchy of clusters from protein interaction data.