Ontology-driven integrative analysis of omics data through Onassis

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Public omics repositories offer valuable data for novel research questions and experimental complementation.
• Data reuse is hindered by challenges in sample identification and organization, despite metadata standards.
• Natural Language Processing (NLP) and biomedical ontologies can address these data organization challenges.

Purpose of the Study:

• To introduce Onassis, an R package designed to simplify the organization and analysis of large-scale omics datasets.
• To leverage NLP and biomedical ontologies for accurate sample annotation and dataset organization.
• To facilitate semantically aware integrative analysis of omics data from public repositories.

Main Methods:

• Development of the Onassis R package within the Bioconductor environment.
• Integration of Natural Language Processing (NLP) tools for text analysis.
• Application of biomedical ontologies for sample annotation and semantic similarity measures for hierarchical dataset organization.

Main Results:

• Onassis effectively simplifies the association of repository samples with ontology-based annotations.
• The package enables hierarchical organization of datasets based on semantic similarity.
• Demonstrated utility across diverse omics datasets, including gene expression, histone marks, and DNA methylation.

Conclusions:

• Onassis leverages NLP and biomedical ontologies within the R statistical framework to identify, relate, and analyze public omics datasets.
• The tool enhances the integrative analysis of various omics data types.
• Onassis significantly facilitates the reuse and interpretation of large-scale omics data.