Self-Attention Neural Network for Protein Structure Prediction

Solution Overview

Problem

Current protein structure prediction methods are computationally intensive and often require extensive resources, relying on iterative search processes and hand-crafted feature engineering, which can be time-consuming and less accurate compared to the proposed system.

Innovation Solution

A system utilizing a pair embedding neural network with self-attention layers to process multiple sequence alignments, generating pair embeddings that are then enriched to predict protein structures efficiently, using a folding neural network to determine structure parameters such as atomic coordinates and backbone torsion angles.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If iterative search processes and hand-crafted feature engineering are used for protein structure prediction, then comprehensive analysis can be performed, but computational resources and time consumption increase significantly

Engineering Contradiction:

Improveprediction accuracyVSAvoidprediction time

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces traditional mechanical iterative search processes with a neural network-based system that uses self-attention mechanisms to directly predict protein structures from amino acid sequences, eliminating the need for time-consuming iterative optimization while maintaining prediction accuracy

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent transforms the prediction approach by changing from hand-crafted features to learned embeddings, where the neural network automatically learns relevant features from multiple sequence alignments, fundamentally altering the parameter representation and reducing manual feature engineering requirements

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If iterative search processes and hand-crafted feature engineering are used for protein structure prediction, then comprehensive analysis can be performed, but computational resources and time consumption increase significantly

Engineering Contradiction:

Improveprediction accuracyVSAvoidcomputational resources

Core Design Contradiction:

Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent replaces traditional mechanical iterative search processes with a neural network-based system that uses self-attention mechanisms to directly predict protein structures from amino acid sequences, eliminating the need for time-consuming iterative optimization while maintaining prediction accuracy

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent segments the protein prediction problem into independent pairwise amino acid relationship predictions, where the self-attention mechanism processes pairs of amino acids separately and aggregates results, reducing overall computational complexity compared to analyzing the entire protein structure globally

Inventive Principle:

Principle #1Segmentation

3Manufacturing precision

If conventional protein structure prediction methods are used, then structure parameters can be determined, but the process is time-consuming and computationally intensive

Engineering Contradiction:

Improvestructure prediction accuracyVSAvoidprediction efficiency

Core Design Contradiction:

Manufacturing precisionVSProductivity

Solution Approach 1:

The patent replaces traditional mechanical iterative search processes with a neural network-based system that uses self-attention mechanisms to directly predict protein structures from amino acid sequences, eliminating the need for time-consuming iterative optimization while maintaining prediction accuracy

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent performs preliminary processing by generating embeddings from multiple sequence alignments before the main prediction step, preparing the data in advance in a way that enables faster and more accurate structure prediction in the subsequent neural network processing stage

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20240412809A1Protein Structure Prediction from Amino Acid Sequences Using Self-Attention Neural Networks

Publication Date: 2024.12.12 GDM HOLDING LLC

AI summary

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining a predicted structure of a protein that is specified by an amino acid sequence. In one aspect, a method comprises: obtaining a multiple sequence alignment for the protein; determining, from the multiple sequence alignment and for each pair of amino acids in the amino acid sequence of the protein, a respective initial embedding of the pair of amino acids; processing the initial embeddings of the pairs of amino acids using a pair embedding neural network comprising a plurality of self-attention neural network layers to generate a final embedding of each pair of amino acids; and determining the predicted structure of the protein based on the final embedding of each pair of amino acids.