2025/11/11 by Dr Abdul Razak Mohamed Sikkander, Mohamed Sikkander, Dr Abdul Razak, Hala S. Abuelmakarem +3
Biochemistry, Genetics and Molecular Biology · #Artificial intelligence #CRISPR Cas9 #CRISPR and Genetic Engineering #Deep learning #Genome editing #Guide RNA optimization #Off target prediction #RNA and protein synthesis mechanisms #RNA regulation and disease
paper · doi:10.71886/bioem.2025.1224242
openalex publication_date 2025/11/11 · openalex created_date 2025/12/21 · openalex updated_date 2026/07/01
Genome‑editing technologies such as CRISPR‑Cas9 have transformed biology and medicine by enabling precise modifications of DNA sequences. Yet key challenges remain: accurate guide‑RNA (gRNA) design, minimizing off‑target effects, optimizing editing efficiency, and tailoring editors to specific cell types or organisms. Artificial intelligence (AI) offers powerful methods—machine learning, deep learning, ensemble models—to learn from large volumes of genomic, epigenomic and experimental editing‑outcome data and thereby design and optimize genome‑editing reagents. This paper reviews the development of AI‑driven genome‑edit tools, presents a hypothetical evaluation framework, and reports results from simulated datasets comparing baseline heuristics versus AI‑augmented design. In our simulation, AI models improved predicted on‑target efficiency by ~25 % and reduced predicted off‑target risk by ~30 % relative to standard rule‑based design. Tabulated results illustrate improvements in gRNA ranking, editor variant selection, and delivery‑vector prediction. We discuss methodological steps: feature engineering (sequence context, chromatin accessibility, cleavage kinetics), model architecture (CNNs, transformer models, ensemble learning), training/validation workflows and deployment considerations (interpretability, regulatory constraints, dataset bias). Limitations include biased training data, cell‑type specificity, delivery challenges, and ethical oversight. Future perspectives emphasize foundation models for editing‑protein design, active‑learning from screening experiments, personalized editing prescriptions, and AI‑augmented clinical pipelines. In conclusion, AI‑powered design and optimisation of genome‑editing tools is poised to accelerate therapeutic, agricultural and synthetic‑biology applications—provided that robust datasets, interpretability and ethical frameworks are in place.