MCAM: multiple clustering analysis methodology for deriving hypotheses and insights from high-throughput proteomic...

Area of Science:

• Proteomics
• Computational Biology
• Systems Biology

Background:

• Proteomic technologies generate vast data on protein states, like phosphorylation.
• Understanding the function of these modifications and responsible enzymes remains a challenge.
• Existing clustering methods often rely on specific assumptions and expert knowledge.

Purpose of the Study:

• To develop a computational framework to infer functional and regulatory meaning of protein states in cell signaling networks.
• To overcome limitations of traditional clustering techniques in biological data analysis.
• To enhance the interpretation of complex proteomic datasets.

Main Methods:

• Developed Multiple Clustering Analysis Methodology (MCAM), a computational framework.
• MCAM uses diverse data transformations, distance metrics, set sizes, and clustering algorithms combinatorially.
• Evaluated clustering sets based on biological insights derived from metadata enrichment (protein functions, kinase substrates, sequence motifs).

Main Results:

• Applied MCAM to ERBB network phosphorylation data, revealing relationships between parameters and biological meaning.
• Generated biological predictions regarding the ERBB network.
• Compared independent and overlapping phosphoproteomic datasets for the ERBB network, highlighting differences based on ligand stimulation and HER2 expression.

Conclusions:

• MCAM is a broadly applicable approach for analyzing proteomic data.
• The method aids in understanding molecular networks across various biological problems.
• MCAM enhances the interpretation of dynamic phosphorylation measurements and phosphoproteomic datasets.