Automated ontological gene annotation for computing disease similarity

Area of Science:

• Genomics
• Bioinformatics
• Medical Informatics

Background:

• Accurate gene-disease association is crucial for clinical diagnosis and drug discovery.
• Existing methods primarily focus on gene expression or genetic data, with limited exploration of protein sequence data for disease terminology mapping.
• The Disease Ontology (DO) provides a valuable framework for organizing disease information.

Purpose of the Study:

• To augment the Disease Ontology (DO) with protein sequence data for automated gene-disease association.
• To improve the accuracy and scope of gene-disease annotations.
• To develop a method for measuring disease similarity based on protein co-occurrence and ontology structure.

Main Methods:

• Augmenting the Disease Ontology (DO) using protein sequence data from Swissprot records.
• Automated annotation of Swissprot records against the augmented DO.
• Measuring disease similarity by analyzing co-occurrence of annotations among proteins and leveraging the hierarchical structure of DO.

Main Results:

• Achieved a 36.1% increase in gene-disease associations compared to the original DO.
• Successfully mapped protein sequence data to disease terminologies.
• Enabled measurement of disease similarity, facilitating the identification of related diseases and potential novel relationships.

Conclusions:

• Augmenting the Disease Ontology with protein sequence data significantly enhances gene-disease associations.
• The developed method provides a robust approach for automated annotation and disease similarity measurement.
• This work has the potential to accelerate clinical diagnosis, drug discovery, and the identification of novel disease connections.