Optical Flow Binding Forecasting for Faster Molecular Dynamics

Solution Overview

Problem

Current molecular dynamic simulations for drug-protein interactions are time-consuming and resource-intensive, spending significant time on unstable binding states, necessitating a method to identify stable binding states early in the simulation to accelerate drug discovery.

Innovation Solution

A computer-implemented method using optical flow analysis and machine learning to predict stable binding states by subsampling data, rendering images from multiple viewpoints, and analyzing atomic displacements to predict binding outcomes within the first few frames of a simulation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If molecular dynamic simulations are performed to predict ligand-protein binding, then binding accuracy is improved, but computing time and resources increase significantly

Engineering Contradiction:

Improvebinding accuracyVSAvoidcomputing time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies preliminary action by using optical flow analysis and machine learning models to predict binding outcomes before completing the full MD simulation. The system analyzes atomic displacements and binding state outcomes from early simulation frames to predict whether a ligand-protein complex will form stable binding, allowing researchers to identify promising candidates early and avoid wasting computational resources on unlikely candidates.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces the traditional mechanical MD simulation approach with a hybrid system that uses optical flow computation and machine learning classification. Instead of relying solely on extensive atomic-level dynamics simulations, the system substitutes parts of the process with image-based optical flow analysis and ML-based binding state prediction, significantly reducing computational time while maintaining accuracy.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

2Measurement precision

If molecular dynamic simulations are performed to predict ligand-protein binding, then binding accuracy is improved, but computing resources increase significantly

Engineering Contradiction:

Improvebinding accuracyVSAvoidcomputing resources

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent applies preliminary action by using optical flow analysis and machine learning models to predict binding outcomes before completing the full MD simulation. The system analyzes atomic displacements and binding state outcomes from early simulation frames to predict whether a ligand-protein complex will form stable binding, allowing researchers to identify promising candidates early and avoid wasting computational resources on unlikely candidates.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent replaces the traditional mechanical MD simulation approach with a hybrid system that uses optical flow computation and machine learning classification. Instead of relying solely on extensive atomic-level dynamics simulations, the system substitutes parts of the process with image-based optical flow analysis and ML-based binding state prediction, significantly reducing computational time while maintaining accuracy.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

3Reliability

If extensive molecular dynamic simulations are performed, then reliable binding state identification is achieved, but simulation time increases

Engineering Contradiction:

Improvebinding state identification reliabilityVSAvoidsimulation time

Core Design Contradiction:

ReliabilityVSLoss of time

Solution Approach 1:

The patent implements feedback by continuously monitoring optical flow patterns and binding state outcomes during the MD simulation and using this information to update the machine learning model's predictions. The system provides real-time feedback on whether the ligand-protein complex is evolving toward a stable binding state, allowing dynamic adjustment of simulation parameters and early identification of reliable binding states without requiring extensive simulation time.

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS12387814B2Optical flow based forecasting of binding states in molecular dynamic simulations

Publication Date: 2025.08.12 INTERNATIONAL BUSINESS MACHINE CORPORATION

AI summary

A computer-implemented method for executing a computation task in a molecular dynamic simulation includes identifying a bonding target on a ligand; constructing a protein structure; rendering an image of the ligand; subsampling data pertaining to the constructed protein structure and the image of the ligand at a particular frequency; rendering a two-dimensional image of the constructed protein structure relative to the ligand from a plurality of viewpoints; computing optical flows of the protein structure relative to the ligand based on the two-dimensional image; analyzing the optical flows to determine a displacement of atoms; simulating a binding state outcome of the protein structure relative to the ligand for each of the plurality of viewpoints; and predicting a probability of the protein structure binding with the ligand, based on the predicted binding state outcome for each of the plurality of viewpoints.