Chemical Structure Encoding for Accurate Polymer Similarity Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for evaluating the similarity of high-molecular compounds have low accuracy due to inadequate methods for decomposing and encoding chemical structural formulas.
Innovation Solution
A non-transitory computer-readable storage medium storing an information processing program that divides the sequence of a compound's rational formula into character strings and branch symbols, assigns compression codes using group and branch dictionaries, and generates coded sequences to improve similarity evaluation accuracy by embedding groups in a Poincare space.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing techniques (amino acid sequence search or vector-based counting) are used to evaluate similarity of high-molecular compounds, then the evaluation process is simple, but the accuracy of similarity evaluation is low
Solution Approach 1:
The patent segments the chemical structural formula into a sequence representation that can be divided into minimum units (character strings) and branch symbols. This segmentation enables precise encoding of compound structures while maintaining systematic organization, resolving the contradiction between evaluation accuracy and system complexity.
Solution Approach 2:
The patent introduces a group dictionary and a group primary structure dictionary as intermediary components between the raw chemical structure and the similarity evaluation process. These dictionaries serve as mediators that systematically map structural features to coded representations, improving accuracy without proportionally increasing overall system complexity.
2Measurement precision
If a detailed encoding system with group dictionary and group primary structure dictionary is implemented, then similarity evaluation accuracy improves, but the system complexity increases
Solution Approach 1:
The encoding system is segmented into two distinct dictionaries: a group dictionary for basic functional groups and a group primary structure dictionary for more complex structures. This segmentation allows the system to handle different levels of structural complexity separately, improving evaluation accuracy while organizing complexity in a manageable, modular way.
Solution Approach 2:
The patent transitions from evaluating compounds in a single dimension (either amino acid sequences or simple vector counts) to a multi-dimensional approach by incorporating both group-level and primary structure-level encoding. This dimensional expansion enables more nuanced similarity evaluation while structuring the complexity across different encoding layers.
Data Source
AI summary
A computer-readable storage medium storing a program for causing a computer to perform processing including: dividing a sequence indicating a rational formula of a compound, into a character string of a minimum unit of the sequence and a branch symbol indicating a branched portion of the compound; generating a first coded sequence by using a group dictionary indicating a relationship between the sequence and the compression code, the generating including assigning a compression code to the character string of the minimum unit, and assigning the compression code according to a type of the branched portion to the branch symbol; and generating a second coded sequence by using a primary structure dictionary indicating a relationship between a group primary structure of the sequence and the compression code, the generating of the second coded sequence including encoding the compression code in the first coded sequence in units of the group primary structure.