High-Density Guide RNA Tiling and Machine Learning for Designing CRISPR Interference in Synechococcus sp. PCC 7002

Area of Science:

• Microbiology
• Molecular Biology
• Synthetic Biology

Background:

• CRISPR interference (CRISPRi) is established in Synechococcus sp. PCC 7002 for gene regulation.
• However, specific design principles for optimizing guide RNA (gRNA) effectiveness are not well understood.

Purpose of the Study:

• To systematically evaluate features influencing gRNA efficiency in Synechococcus sp. PCC 7002.
• To develop predictive models for gRNA design to tune gene expression.

Main Methods:

• Constructed 76 Synechococcus sp. PCC 7002 strains with various gRNAs targeting reporter systems.
• Performed correlation analysis to identify significant gRNA design features.
• Applied machine learning algorithms, including Random Forest, to predict gRNA effectiveness.

Main Results:

• Identified critical gRNA design features: position relative to start codon, GC content, PAM site, minimum free energy, and DNA strand.
• Observed unexpected gene activation with promoter-targeting gRNAs and enhanced repression with terminator-targeting gRNAs.
• Random Forest model demonstrated the highest accuracy in predicting gRNA effectiveness.

Conclusions:

• High-density gRNA data combined with machine learning significantly improves the prediction of gRNA efficiency.
• These findings provide actionable insights for designing effective gRNAs to precisely control gene expression in Synechococcus sp. PCC 7002.