Gilda: biomedical entity text normalization with machine-learned disambiguation as a service

Area of Science:

• Biomedical Informatics
• Computational Biology

Background:

• Biomedical research relies on accurate identification of entities like genes, proteins, and diseases.
• Ambiguity in naming and synonyms across different databases poses a significant challenge for data integration and analysis.

Purpose of the Study:

• To develop and present Gilda, a software tool and web service for robust biomedical entity recognition.
• To improve the accuracy of matching names and synonyms across diverse biomedical ontologies.

Main Methods:

• Implementation of a scored string matching algorithm for biomedical entity recognition.
• Integration of machine-learned disambiguation models utilizing contextual information.
• Inclusion of species-prioritization functionality to resolve ambiguity.

Main Results:

• Gilda effectively matches names and synonyms for genes, proteins, small molecules, biological processes, and diseases.
• Machine-learned disambiguation models enhance accuracy by considering surrounding text.
• The tool supports species-specific entity resolution.

Conclusions:

• Gilda provides a powerful and accurate solution for grounding biomedical concepts.
• The software facilitates more reliable data integration and analysis in life sciences.
• Gilda is accessible as a web service with open-source code and documentation.