Massively parallel CRISPR off-target detection enables rapid off-target prediction model building

Area of Science:

• Molecular Biology
• Genetics
• Biotechnology

Background:

• CRISPR genome editing offers significant clinical potential but faces challenges with off-target effects.
• Accurate detection of these off-target mutations is crucial for safe clinical translation.

Purpose of the Study:

• To develop a sensitive and specific method for detecting low-frequency off-target mutations in CRISPR genome editing.
• To establish a high-throughput platform for selecting optimal guide RNAs and characterizing novel CRISPR nucleases.

Main Methods:

• Developed adaptor-mediated off-target identification by sequencing (AID-seq) for comprehensive off-target detection.
• Implemented a pooled strategy with AID-seq to analyze multiple guide RNAs simultaneously.
• Utilized deep learning (CRISPR-Net) to build an off-target prediction model.

Main Results:

• AID-seq demonstrated high sensitivity and specificity in detecting off-target mutations from various CRISPR nucleases (Cas9, Cas12a).
• The pooled strategy successfully screened efficient and safe targets for antiviral therapy and profiled a new CRISPR (FrCas9).
• The CRISPR-Net model achieved high performance (AUROC = 0.97, AUPRC = 0.29) in predicting off-target effects.

Conclusions:

• AID-seq represents the most sensitive and specific in vitro off-target detection method currently available.
• The pooled AID-seq strategy provides a rapid, high-throughput platform for optimizing CRISPR applications and discovering new systems.