The compositional structure of Gene Ontology terms

Area of Science:

• Bioinformatics
• Computational Biology
• Ontology Engineering

Background:

• The Gene Ontology (GO) is a critical resource for annotating gene and protein functions.
• Understanding the structural relationships within GO terms is essential for effective data mining and interpretation.
• Previous analyses have not fully explored the prevalence and implications of substring relationships among GO terms.

Purpose of the Study:

• To quantify the extent of substring relationships within the Gene Ontology term names.
• To investigate the semantic significance of recurring substrings and derivational phrases.
• To propose methods for leveraging these linguistic patterns to improve GO's computational accessibility and data representation.

Main Methods:

• Algorithmic analysis of all Gene Ontology term names to identify substring occurrences.
• Statistical evaluation of the frequency and distribution of specific substrings.
• Examination of the correlation between substring presence and established derivational relationships.

Main Results:

• A significant proportion (65.3%) of Gene Ontology terms contain another GO term as a proper substring.
• Recurring substrings, such as 'regulation of', frequently appear in consistent subtrees, indicating semantic relationships.
• These substring patterns often reflect a clear derivational structure between related GO terms.

Conclusions:

• Substring relationships are a prevalent structural feature of the Gene Ontology.
• Frequently occurring substrings act as meaningful linguistic markers, revealing semantic connections.
• Exploiting these substring phenomena can enhance computational accessibility, enrich data representation, and aid natural language processing tasks related to GO.