Clinical trials have begun using CRISPR/Cas genome editing to treat certain diseases, making safety a top priority. The CRISPR/Cas system is an efficient and versatile approach to gene editing, using a single guide RNA (sgRNA) to induce precise DNA modifications. However, screening sgRNAs for efficacy and specificity is time-consuming and resource-intensive. To streamline this process, we present Navitas/Optimus, a novel computational tool that predicts the optimal sgRNA for a given target sequence. Navitas features a unique approach of not differentiating between on- and off-target activity, and employs a combination of Machine Learning models, including XGBoost, Linear Regression, Random Forest, and Feedforward Neural Network (FNN), with FNN performing the best overall. The results of Navitas, which utilized FNN, are further optimized by data augmentation. Optimus, an Evolutionary Algorithm (EA), optimizes the prediction of the optimal sgRNA given a target DNA sequence using Navitas as its fitness function. Navitas/Optimus has the potential to contribute to the field of genome editing by offering a novel way of predicting the optimal sgRNA for a given target sequence.
Article ID: 2023S7
Publisher: Canadian Artificial Intelligence Association