AMBIT RESTful web services: an implementation of the OpenTox application programming interface.

Area of Science:

• Computational toxicology
• cheminformatics
• bioinformatics

Background:

• The Open Source Predictive Toxicology Framework (OpenTox) project aimed to unify access to toxicity data and predictive models.
• Existing implementations lacked a standardized approach, hindering interoperability.

Purpose of the Study:

• To present the AMBIT web services package as a Representational State Transfer (REST) architecture-compliant implementation of the OpenTox API.
• To facilitate unified access to toxicity data, predictive models, and validation procedures.
• To enable the creation of Quantitative Structure-Activity Relationship (QSAR) models and integrate toxicological hazard prediction modules.

Main Methods:

• Developed using REST architecture principles and adhering to the OpenTox API.
• Utilized an information model based on OWL-DL ontology and linked data principles (Resource Description Framework - RDF).
• Integrated various algorithms for descriptor calculation, machine learning, similarity searching, and toxicological hazard prediction (e.g., Toxtree modules).

Main Results:

• The AMBIT package provides a unified interface for accessing and processing chemical data and predictive models.
• Datasets with chemical structures and properties are automatically available online in multiple formats.
• The service simplifies running predictions and sharing datasets/models without software installation.
• Enabled seamless interoperability within applications and across distributed services.

Conclusions:

• The AMBIT web services package offers a standardized, interoperable solution for predictive toxicology.
• It facilitates easier access to toxicity data, model building, and prediction execution.
• The framework supports the development of web mashups, end-user applications, and integration into existing workflows.