Prediction of Base Editing Efficiencies and Outcomes Using DeepABE and DeepCBE

Area of Science:

• Genetics and Genomics
• Bioinformatics
• Molecular Biology

Background:

• Adenine base editors (ABEs) and cytosine base editors (CBEs) are crucial for introducing disease-relevant point mutations.
• Challenges in base editing include low efficiency and multiple target nucleotides within the editing window.
• Previous methods for evaluating editing outcomes are time-consuming and labor-intensive.

Purpose of the Study:

• To develop deep learning models (DeepABE and DeepCBE) for in silico prediction of base editing efficiencies and outcome frequencies.
• To provide a user-friendly web tool (DeepBaseEditor) for accurate determination of specific target nucleotides for ABE and CBE editing.

Main Methods:

• Development of deep learning models, DeepABE and DeepCBE, utilizing computational approaches.
• Implementation of these models into an accessible online web tool, DeepBaseEditor.
• Validation of the models' accuracy in predicting base editing outcomes.

Main Results:

• DeepABE and DeepCBE accurately predict base editing efficiencies and outcome frequencies at target DNA sites.
• The DeepBaseEditor web tool facilitates the in silico determination of optimal target nucleotides for ABE and CBE editing.
• Computational prediction significantly reduces the need for extensive experimental evaluation.

Conclusions:

• Deep learning models offer a powerful and efficient approach to predict base editing outcomes.
• The DeepBaseEditor tool streamlines the process of designing gene editing experiments.
• This approach enhances the precision and efficiency of introducing specific point mutations for research and therapeutic applications.