ComplexFinder

ComplexFinder analyzes native protein complex fractionation experiments to identify and characterize protein complexes from Blue-Native and Size-Exclusion Chromatography (SEC) fractionation coupled to LC-MS/MS and other separation techniques that produce co-elution signal profiles.

Key Features:

• Complexome Profiling Integration: Integrates complexome profiling with high-resolution mass spectrometry to identify protein complexes in an unbiased manner.
• Machine Learning-Based Prediction: Employs machine-learning to predict novel protein-protein interactions using various measures of distance between signal profiles.
• Peak-Centric Signal Profile Representation: Represents each protein's signal profile as an ensemble of peak-like models to enable calculation of local similarities and peak-centric comparisons across conditions.
• Protein Connectivity Network Construction: Constructs protein connectivity networks from predicted protein-protein interactions and assembles proteins into macromolecular complexes using peak-centric information.
• Support for Multiple Quantification Strategies: Supports LC-MS/MS quantification strategies including label-free, SILAC, TMT, and pulseSILAC.
• Compatibility with Multiple Fractionation Techniques: Applicable to Blue-Native and SEC fractionation and to any separation technique that yields co-elution signal profiles.

Scientific Applications:

• Protein Complex Identification: Identification and composition estimation of macromolecular protein complexes from fractionation data.
• Protein Distribution Quantification: Quantification of the distribution of individual proteins across different complexes and biological conditions.
• Comparative Complexome Analysis: Peak-centric comparisons across biological conditions to detect remodeling of complex composition.
• High-Resolution LC-MS/MS Data Analysis: Analysis of high-resolution LC-MS/MS data from Blue-Native and SEC fractionation, addressing increased data acquisition rates from advanced instrumentation.

Methodology:

Computational steps include representing signal profiles as ensembles of peak-like models, calculating local similarities and performing peak-centric comparisons, assembling potential complexes via correlation analysis, using machine-learning with distance measures between signal profiles to predict protein-protein interactions, and constructing protein connectivity networks to assemble macromolecular complexes.