SummArIzeR: simplifying cross-database enrichment result clustering and annotation via large language models

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Enrichment analysis across multiple databases leads to redundant terms, complicating biological data interpretation.
• Overlapping terms in enrichment analysis hinder fast, intuitive interpretation and comparison across conditions.
• Existing tools lack efficient methods for clustering and annotating complex enrichment results.

Purpose of the Study:

• To develop an R package, SummArIzeR, for clustering and annotating enrichment results across multiple databases.
• To enable fast, intuitive interpretation and comparison of biological data across multiple conditions.
• To facilitate the annotation of enrichment clusters using large-language models.

Main Methods:

• SummArIzeR clusters enrichment results based on shared genes.
• A pooled p-value is calculated for each cluster.
• Large-language models are utilized for cluster annotation.
• The package provides easily interpretable visualizations of the results.

Main Results:

• SummArIzeR offers unbiased and fast cluster annotation powered by large language models.
• The package achieves clustering comparable to manual curation.
• SummArIzeR provides superior grouping of enrichment results based on shared genes.

Conclusions:

• SummArIzeR enhances the interpretation of biological enrichment analysis.
• The R package offers an efficient and intuitive approach to managing complex enrichment data.
• SummArIzeR is available as an open-source R package with a user manual on GitHub.