clusterMaker2

clusterMaker2 performs multi-algorithm clustering, node ranking, and dimensionality-reduction analyses on molecular interaction networks and high-dimensional omics data to identify and rank biologically relevant modules.

Key Features:

• Integration with Cytoscape: Operates on molecular interaction networks and biological pathways within the Cytoscape environment.
• Expanded Algorithm Suite: Includes multiple algorithm categories, notably node ranking and dimensionality reduction.
• Node Ranking Algorithms: Prioritizes or scores nodes such as genes or proteins within networks based on specified criteria.
• Dimensionality Reduction (UMAP): Implements UMAP (Uniform Manifold Approximation and Projection) for lowering data dimensionality and visualizing structure.
• Fuzzy Clustering: Assigns elements to multiple clusters with graded membership to capture overlapping biological signals.
• Cluster Ranking: Ranks clusters by significance or relevance to help identify biologically meaningful groups.
• Leiden Clustering: Segments large networks into communities using the Leiden algorithm.
• Hierarchical Clustering: Produces hierarchical representations of expression datasets.
• Cytoscape Jobs API Integration: Implements many algorithms via the Cytoscape Jobs API to execute remote jobs.

Scientific Applications:

• Genomics: Identification and ranking of gene modules from expression and interaction data.
• Proteomics: Detection of protein complexes and functional groupings in protein-protein interaction networks.
• Systems Biology: Analysis of network modules and pathway-level organization in biological systems.
• Single-cell Transcriptomics: Dimensionality reduction and clustering to reveal cell populations and expression patterns.
• Network-based Functional Module Identification: Integration with PPI resources such as STRING to find modules related to specific conditions, exemplified by yeast heat shock response analysis.

Methodology:

Applied Leiden clustering, hierarchical clustering, UMAP dimensionality reduction, and fuzzy clustering to yeast heat shock expression data combined with the yeast protein-protein interaction network from STRING, with many algorithms executed via the Cytoscape Jobs API.