Scalable Query Answering Under Uncertainty to Neuroscientific Ontological Knowledge: The NeuroLang Approach

Area of Science:

• Neuroscience
• Computational Neuroscience
• Data Science

Background:

• Growing availability of neuroscience datasets due to technological advances.
• Existing ontological knowledge for brain areas and activation patterns.
• Inherent uncertainty in brain imaging data (voxel mapping).
• Lack of a unified framework for accessing heterogeneous neuroimaging data under uncertainty.

Purpose of the Study:

• To introduce NeuroLang, a novel probabilistic language for neuroscience research.
• To provide a unified framework for integrating heterogeneous data and ontologies.
• To enable tractable query answering over large, uncertain brain datasets.
• To map cognitive domains to brain regions formally and promote reproducible research.

Main Methods:

• Development of NeuroLang, a probabilistic language based on first-order logic with existential rules.
• Integration of probabilistic uncertainty and ontological knowledge under the open-world assumption.
• Inclusion of built-in mechanisms for tractable query answering.
• Presentation of a general query answering architecture.

Main Results:

• NeuroLang provides a unified framework for heterogeneous neuroscience data.
• The language handles probabilistic uncertainty inherent in brain scans.
• It facilitates the integration of ontologies and formal mapping of cognitive domains to brain regions.
• Demonstrated applicability through real-world use cases.

Conclusions:

• NeuroLang offers a solution for accessing and querying complex, uncertain neuroscience data.
• The framework promotes shareable and highly reproducible research in neuroscience.
• It addresses limitations of current ad hoc tools and limited query languages.