CLNN-loop: a deep learning model to predict CTCF-mediated chromatin loops in the different cell lines and...

Area of Science:

• Genomics and Bioinformatics
• Computational Biology
• Molecular Biology

Background:

• Three-dimensional (3D) genome organization plays a critical role in gene regulation and disease.
• CTCF-mediated chromatin loops are essential components of 3D genome structure, but experimental detection is costly and time-consuming.
• Existing sequence-based computational methods for predicting chromatin loops suffer from low performance and poor generalization.

Purpose of the Study:

• To develop a novel deep learning model for accurate prediction of chromatin loops.
• To improve the prediction of chromatin loops across different cell lines and CTCF-binding sites (CBS) pair types.
• To identify key sequence-based features that drive chromatin loop formation.

Main Methods:

• Proposed CLNN-loop, a novel deep learning model.
• Fused multiple sequence-based features for enhanced prediction.
• Utilized the SHAP framework for model interpretability.

Main Results:

• CLNN-loop demonstrated satisfactory performance and superiority over existing methods in predicting chromatin loops.
• The model achieved effective predictions across diverse cell lines and CBS pair types.
• SHAP analysis revealed that CTCF motif and sequence conservation are crucial indicators for chromatin loops.

Conclusions:

• CLNN-loop offers a powerful and accurate computational approach for predicting chromatin loops.
• The findings highlight the importance of sequence-based features, particularly CTCF motifs and conservation, in 3D genome organization.
• The developed model and its webserver provide valuable resources for genomic research.