CRISPR-GEM: A Novel Machine Learning Model for CRISPR Genetic Target Discovery and Evaluation

Area of Science:

• Biotechnology
• Genomics
• Computational Biology

Background:

• CRISPR gene editing offers precise control over gene expression for cell therapies.
• Effective CRISPR strategies require careful selection of target genes based on their role in gene regulatory networks (GRNs).
• Current methods for decoding GRNs are limited to single cell types and transcription factors, restricting their use in CRISPR applications.

Purpose of the Study:

• To develop a novel machine learning model, CRISPR-GEM, for predicting the downstream effects of CRISPR gene editing.
• To enable informed selection of optimal CRISPR target genes for enhanced cell therapy development.
• To overcome limitations of existing GRN analysis techniques for CRISPR strategy optimization.

Main Methods:

• CRISPR-GEM utilizes a multilayer perceptron (MLP) based synthetic GRN to predict gene expression changes.
• Input and output nodes are defined as differentially expressed genes between experimental and target cell types.
• The model is trained using a black-box approach to learn regulatory relationships and predict gene expression, followed by CRISPR-mimetic perturbations to score candidate genes.

Main Results:

• CRISPR-GEM accurately predicts the downstream effects of CRISPR gene editing on gene expression.
• The model identifies top-scoring genes that effectively modulate GRNs to achieve desired cell phenotypes.
• This represents the first machine learning model specifically designed for predicting optimal CRISPR target genes.

Conclusions:

• CRISPR-GEM provides a powerful tool for enhancing CRISPR-based cell therapies.
• The model facilitates the selection of optimal CRISPR targets by predicting their impact on cellular phenotype.
• This approach significantly advances the development of precise and effective gene editing strategies for therapeutic applications.