IPPI

IPPI infers protein-protein interactions and their probabilities from experimental interaction-strength data using a probabilistic model formulated as a linear programming problem.

Key Features:

• Probabilistic Model-Based Inference: Employs a probabilistic model that uses experimental interaction-strength data and interaction ratios and minimizes discrepancies between observed interaction ratios and predicted probabilities during training.
• Linear Programming Framework: Formulates the inference problem as a linear programming optimization to derive predicted interaction probabilities from available experimental data.
• Comparative Performance: Demonstrates robust performance in comparisons with association, EM (Expectation-Maximization), and SVM-based approaches, particularly for numerical interaction data while remaining competitive on binary datasets.
• Data Transformation Tools: Includes programs to convert input experimental data into the format required by the linear programming model.

Scientific Applications:

• PPI Network Modeling: Enables inference of quantitative protein-protein interaction networks from multiple experiments per protein pair.
• Cellular Process Analysis: Supports construction of interaction-strength-informed models to study cellular processes.
• Disease Mechanism and Target Analysis: Facilitates analysis of disease mechanisms and the identification of potential therapeutic targets through refined PPI inferences.

Methodology:

Uses a probabilistic model that leverages experimental interaction-strength data and interaction ratios, trains by minimizing discrepancies between observed ratios and predicted probabilities, and formulates the inference as a linear programming problem.