AI Compound Fragment Modification for Selective Drug Candidate Design
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing AI-based molecular structure designing methods struggle to selectively modify only a part of a given molecular structure, leading to inefficiencies in generating new drug candidates, as they often generate unwanted structures and fail to utilize important partial structures of potent materials effectively.
Innovation Solution
A method and apparatus that utilize AI to divide a source molecular structure into partial structures, allowing selective modification and application of only a partial structure to generate new materials with similar action, considering synthesis difficulty, similarity to existing structures, and compatibility with target proteins.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing AI-based molecular structure designing methods generate a new complete molecular structure, then automation and speed are improved, but the ability to selectively modify only a part of the structure is lost and many unwanted structures are generated
Solution Approach 1:
The molecular structure is divided into multiple partial structures (fragments), allowing the AI to selectively modify only the desired portion while keeping other parts fixed. This segmentation enables targeted modification rather than generating entirely new structures, thus improving both selectivity and efficiency.
Solution Approach 2:
The desired partial structure from a potent material is extracted and fixed, while only the remaining portions are subjected to AI-based modification. This extraction approach allows the important functional parts to be preserved while still enabling automated design of the rest of the molecule.
2Extent of automation
If existing AI-based methods generate complete new molecular structures, then automation is improved, but efficiency is greatly reduced due to generation of unwanted structures
Solution Approach 1:
Different parts of the molecular structure are treated differently - some parts are fixed (important functional groups) while others are modified (less critical regions). This local differentiation allows automation to focus only on the portions that need modification, reducing wasted computational effort on unwanted structures.
Solution Approach 2:
Instead of completely regenerating the entire molecular structure, the AI performs partial modification only on the necessary portions. This partial action approach maintains automation while significantly improving efficiency by avoiding unnecessary generation and rejection of unwanted structures.
3Loss of information
If partial structures of potent materials are applied to generation of candidate materials, then utilization of important information is improved, but efficiency is greatly reduced without proper fixing methods
Solution Approach 1:
The important partial structures are identified and fixed in advance before the AI-based generation process begins. This preliminary action ensures that critical information from potent materials is preserved and properly utilized, while the AI then efficiently generates variations only in the non-fixed portions.
Solution Approach 2:
By segmenting the molecular structure into fixed and modifiable portions, the system can effectively utilize the important information contained in the fixed partial structures while maintaining high efficiency in generating new candidate materials through automated modification of the variable portions.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
The present disclosure provides a method of generating compound information in a computing apparatus, the method including obtaining a learning model for information associated with partial structures, obtaining information associated with a source molecule that is a target of a partial structure modification, obtaining information associated with a partial structure set including a plurality of partial structures of the source molecule, selecting, from the partial structures included in the partial structure set, a target partial structure to be modified, obtaining, using the learning model, information associated with a modified partial structure corresponding to the target partial structure, and outputting result information in which the target partial structure is replaced by the modified partial structure in the source molecule.