Scientists have developed a novel machine learning approach that enhances the prediction of the efficiency of CRISPR interference (CRISPRi) for specific genes. CRISPRi is a gene-editing tool that blocks gene expression without modifying the DNA sequence. Although widely used to silence gene expression in bacteria, the design rules for CRISPRi experiments have remained poorly defined.
To address this challenge, researchers utilized data integration and artificial intelligence (AI) to train a machine learning model. The team employed multiple genome-wide CRISPRi essentiality screens to enhance the prediction of guide RNA efficiency in the CRISPRi system.
Their findings indicate that gene-specific characteristics significantly influence guide RNA depletion in genome-wide screens. Furthermore, combining data from multiple CRISPRi screens substantially enhances the accuracy of prediction models and provides more reliable estimates of guide RNA efficiency. By improving the understanding of guide RNA efficiency, this study informs the development of precise gene-silencing strategies using CRISPRi.
Led by Lars Barquist, PhD, research group leader at the Würzburg Helmholtz Institute for RNA-based Infection Research (HIRI) and junior professor at the University of Würzburg, the researchers developed a mixed-effect random forest regression model that yields more accurate predictions of guide RNA efficiency. The study authors validated their approach by conducting an independent screen targeting essential bacterial genes, which demonstrated the superiority of their predictions over existing methods.
The study also highlighted a surprising finding that gene expression-related characteristics have a greater impact on CRISPRi depletion in essentiality screens than the guide RNA itself. This insight challenges the previously assumed primary role of guide RNA in CRISPRi efficiency.
The integration of data from multiple experiments was key in building more precise prediction models, overcoming the lack of data as a major limitation for accuracy. The findings of this study lay the groundwork for the development of improved tools to manipulate bacterial gene expression, furthering our understanding of pathogens and aiding in the development of targeted therapeutic interventions.
This study, titled “Improved prediction of bacterial CRISPRi guide efficiency from depletion screens through mixed-effect machine learning and data integration,” was published in Genome Biology.