A tool for CRISPR-Cas9 sgRNA evaluation based on computational models of gene expression

Area of Science:

• Genomics
• Molecular Biology
• Bioinformatics

Background:

• CRISPR gene silencing relies on mutations but often overlooks gene expression impacts.
• Existing tools predict CRISPR knockouts but lack gene expression analysis.
• Accurate prediction of CRISPR outcomes requires integrating gene expression data.

Purpose of the Study:

• To develop a computational tool, EXPosition, for predicting CRISPR outcomes by incorporating gene expression.
• To enhance the accuracy of CRISPR-induced gene knockout predictions.
• To provide a tool that aids in genome editing project efficiency and success.

Main Methods:

• Developed EXPosition, a novel computational tool combining knockout and gene expression prediction models.
• Utilized deep learning to model transcription, splicing, and translation initiation.
• Validated models using 6 datasets, 4 cell types, and ~207k sgRNAs, and the ClinVar dataset.

Main Results:

• EXPosition outperforms existing tools in predicting gene knockouts across diverse datasets and cell types.
• Deep learning models accurately forecast key gene expression steps.
• Gene expression models showed enrichment of pathogenic mutations in high-scoring mutations in ClinVar data.

Conclusions:

• EXPosition enhances genome editing efficiency and accuracy by predicting CRISPR's effects on gene expression.
• The tool offers improved prediction of CRISPR-induced gene silencing.
• EXPosition is publicly available for research use.