Allosteric Path Prediction via Protein Network Graphs

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current technologies face challenges in predicting amino acid residues contributing to allosteric control from three-dimensional structure information of proteins, which is crucial for drug development, especially in allosteric drug development.

Innovation Solution

An allosteric path prediction device and method that generates a network graph based on protein three-dimensional structure information, using vertices for amino acid residues and assigning weights to edges based on interactions, to calculate signal transmission paths and identify important amino acid residues.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If network graph generation and path calculation are performed to predict allosteric paths, then prediction accuracy of amino acid residues is improved, but computational complexity and processing time increase

Engineering Contradiction:
Improveprediction accuracyVSAvoidcomputational complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the protein structure into a network graph where amino acid residues are represented as nodes and interactions as edges. This segmentation transforms the complex protein structure analysis into a manageable graph problem, allowing systematic path calculation while maintaining prediction accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces an evaluation function as an intermediary between the network graph and path calculation. This evaluation function quantifies the quality of potential allosteric paths, enabling the system to efficiently identify relevant paths without exhaustively analyzing all possible paths through the protein structure.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If comprehensive interaction weights are assigned to all edges in the network graph, then prediction reliability is improved, but computational resources and processing time increase

Engineering Contradiction:
Improveprediction reliabilityVSAvoidcomputational resources
Core Design Contradiction:
ReliabilityVSUse of energy by stationary object

Solution Approach 1:

The patent applies local quality by assigning different weight values to different edges in the network graph based on the specific interaction characteristics between amino acid residue pairs. This allows the system to focus computational resources on evaluating locally important interactions while maintaining overall prediction reliability.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent uses parameter changes by varying the weight values assigned to edges based on interaction types (e.g., hydrophobic, hydrogen bonding, electrostatic). This parameter-based approach allows comprehensive evaluation of interactions while enabling efficient computation through weighted prioritization of different interaction strengths.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4546346A1Allosteric path prediction device, allosteric path prediction result acquisition method, allosteric path prediction method, and program
Publication Date: 2025.04.30 TOSHIBA DIGITAL SOLUTIONS CORP
  • EP4546346A1 patent drawingFigure 1
  • EP4546346A1 patent drawingFigure 2
  • EP4546346A1 patent drawingFigure 3

AI summary

An allosteric path prediction device includes a network graph generating unit and a path calculating unit. The network graph generating unit generates, on the basis of three-dimensional structure information of a protein, a network graph which includes vertices corresponding to at least amino acid residues constituting the protein out of the amino acid residues and arbitrary binding substances bound to the protein and in which weights based on interactions between at least the amino acid residues out of the amino acid residues and the arbitrary binding substances are assigned to edges on the basis of three-dimensional structure information of the protein. The path calculating unit calculates a path connecting the vertices on the network graph generated by the network graph generating unit on the basis of an evaluation function based on the weights.