regBase: whole genome base-wise aggregation and functional prediction for human non-coding regulatory variants

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Interpreting genetic variations in the human non-coding genome is crucial for understanding disease causation.
• Existing prediction methods for regulatory variants often show inconsistent performance or are limited in scope.
• There is a need for integrated approaches to comprehensively score non-coding regulatory variants.

Purpose of the Study:

• To develop an integrated resource and predictive models for non-coding regulatory variants.
• To improve the identification of functional, pathogenic, and cancer driver variants.
• To enhance the fine-mapping of causal regulatory variants.

Main Methods:

• Compiled whole-genome base-wise aggregations into the regBase database.
• Developed three composite models based on different causality assumptions.
• Trained models to score functional, pathogenic, and cancer driver non-coding regulatory variants.
• Validated model performance using independent benchmarks.

Main Results:

• The regBase database integrates a large number of prediction scores.
• The developed composite models demonstrate superior and stable performance.
• Successfully fine-mapped causal regulatory variants at specific loci and base-wise resolution.

Conclusions:

• The regBase database and composite models offer a powerful tool for human genetic studies.
• These resources aid in annotation-based causal variant fine-mapping.
• The tools facilitate pathogenic variant discovery and cancer driver mutation identification.