Defining peptides in ChEBI

Area of Science:

• Biochemistry
• Bioinformatics
• Ontology Engineering

Background:

• Modern biochemistry generates extensive chemical data, necessitating efficient knowledge organization.
• Existing ontologies like Chemical Entities of Biological Interest (ChEBI) struggle to keep pace with domain growth through manual classification alone.

Purpose of the Study:

• To propose a novel, comprehensive taxonomy of 67 peptide-related classes for the ChEBI ontology.
• To develop a methodology for automated classification based on formal logical axiomatization.

Main Methods:

• Expanded and refined natural language definitions for peptide classes.
• Formalized definitions using monadic second-order logic (MSOL).
• Developed a translation methodology from MSOL to algorithmic classification.

Main Results:

• Created a new taxonomy with 53 novel peptide classes and refined 14 existing ones.
• Implemented an efficient algorithm for large-scale molecule classification.
• Identified potential inconsistencies within the current ChEBI taxonomy through comparative analysis.

Conclusions:

• The proposed automated classification method enhances ontological precision and consistency.
• This approach facilitates the large-scale organization and validation of chemical knowledge.
• Expert evaluation confirmed the quality of the natural language definitions.