Advancing CRISPR with deep learning: A comprehensive review of models and databases

Area of Science:

• Genetics and Genomics
• Bioinformatics
• Biotechnology

Background:

• CRISPR technology offers powerful targeted genome editing capabilities.
• Off-target effects are a major hurdle limiting the clinical application of CRISPR gene editing therapies.
• Accurate prediction of CRISPR editing efficiency is vital for advancing gene-editing therapies.

Purpose of the Study:

• To review available CRISPR databases for enhancing the prediction of CRISPR on-target and off-target activity.
• To highlight the role of machine learning (ML) and deep learning (DL) in improving CRISPR efficiency predictions.
• To address the need for large, searchable databases to support ML/DL approaches in CRISPR research.

Main Methods:

• Review of existing literature on CRISPR databases.
• Analysis of machine learning and deep learning methodologies for predicting CRISPR activity.
• Discussion of sequence features and their incorporation into prediction models.

Main Results:

• Current prediction accuracy for CRISPR on-target and off-target activity is limited by training data availability.
• Machine learning and deep learning are emerging as leading methods for CRISPR efficiency prediction.
• The development of comprehensive and searchable CRISPR databases is essential for advancing ML/DL prediction tools.

Conclusions:

• Improved prediction of CRISPR efficiency is critical for the safe and effective clinical application of gene-editing therapies.
• Machine learning and deep learning approaches, supported by extensive databases, show significant promise in overcoming CRISPR's off-target challenges.
• Further development and integration of sequence features into DL models are expected to enhance prediction accuracy and clinical utility.