CENTRE: a gradient boosting algorithm for Cell-type-specific ENhancer-Target pREdiction.

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Identifying target promoters of active enhancers is crucial for understanding gene regulation, phenotypes, and diseases.
• Current computational methods for predicting enhancer-gene interactions often require extensive cell-type-specific experimental data or large cohorts, making predictions for unstudied cell types laborious and costly.

Purpose of the Study:

• To develop a machine learning framework (CENTRE) for inferring cell-type-specific enhancer target interactions with minimal experimental input.
• To provide a cost-effective and efficient method for predicting enhancer-gene interactions in diverse cell types.

Main Methods:

• Introduced Cell-specific ENhancer Target pREdiction (CENTRE), a machine learning framework.
• Utilized gene expression and ChIP-seq data for three histone modifications from the cell type of interest.
• Integrated cell-type-agnostic statistics from available datasets to enhance cell-type-specific predictions.

Main Results:

• CENTRE accurately predicts cell-type-specific enhancer target interactions using limited experimental data.
• The framework achieves performance equivalent or superior to existing methods that require extensive experimental data.
• Demonstrated robustness across multiple datasets and cell types.

Conclusions:

• CENTRE offers an efficient and accurate approach to predict cell-type-specific enhancer-gene interactions.
• The framework reduces the need for extensive experimental assays, making it valuable for unstudied cell types.
• CENTRE is available as open-source code, facilitating its adoption and further research.