Artificial Intelligence for Predictive Modeling in CRISPR/Cas9 Gene Editing: a Survey of Methods and Design Strategies
Jay Patel, Dhruvi Patel, Aanal Raval
Abstract
Ongoing developments in genome editing most notably the continued evolution of CRISPR-Cas systems and their orthogonal or modified counterparts have substantively altered both experimental and applied practices in biomedicine, agriculture, and therapeutic design. More recently, the systematic incorporation of artificial intelligence and machine learning methodologies has augmented the specificity, throughput, and explanatory capacity of genome-editing workflows, thereby refining the prediction of on-target efficiencies, the appraisal of off-target liabilities, and the tailoring of molecular therapeutic configurations. The present contribution offers an integrative survey of these computational developments, emphasizing (i) predictive algorithms, (ii) machine-learning and deep-learning frameworks, (iii) data-centric procedural strategies, and (iv) dedicated applications in oncology, neurology, rare-disease research, and precision-medicine contexts. Throughout, we evaluate architectural choices, sequence-encoding representations, and lingering dataset-related biases, while additionally addressing current constraints concerning model interpretability, ethical viability, and the procedural prerequisites for clinical translation. Moreover, we advance a structured taxonomy that organizes AI-mediated genome-editing approaches according to methodological lineage and functional scope, and we delineate extant research lacunae. By combining these elements, we supply a prospective assessment of the means by which artificial intelligence might be further leveraged to support secure, efficacious, and equitably accessible genome engineering outcomes.