BE-dataHIVE

BE-dataHIVE stores over 460,000 base editing gRNA target combinations in a MySQL database to enable analysis of base editing data and prediction of outcomes for practitioners and machine learning researchers.

Key Features:

• Large-scale dataset: The database contains over 460,000 gRNA target combinations compiled from five studies.
• Enriched molecular features: Data are annotated with melting temperatures and energy terms.
• Precomputed machine learning structures: Multiple different data structures for machine learning were computed and are directly available.
• Programmatic access: The database provides an API and Python library for data analysis and outcome prediction.

Scientific Applications:

• Base editing outcome prediction: Supports computational modeling to predict base editing accuracy and efficiency.
• By-stander mutation minimization: Facilitates the design of base editors with reduced unintended mutations for clinical applications.
• Machine learning model training: Provides a diverse, feature-rich, and large dataset to improve the robustness and performance of base editing prediction models.
• Gene editing research: Enables practitioners and researchers to analyze base editing data for precise single-nucleotide transformations.

Methodology:

Data from five studies were compiled into a MySQL database and enriched with melting temperatures and energy terms; multiple data structures for machine learning were computed.