Go to content

From Hack and Tip to AI Precision: The Evolution of CRISPR gRNA Design (Invited) | BioNatura Journal Vol. 2 No. 3 (2025) - Bionatura journal

Ibero-American Journal of Biotechnology and Life Sciences
________________________________________________________________________________
______________________________________________________________________________
Microbiology, Bacteriophages & Antimicrobial Resistance Biomedical Research, Pharmacology & Clinical Case Reports Sustainable Agriculture, Biostimulants & Plant Biotechnology

From Hack and Tip to AI Precision: the Evolution of CRISPR gRNA Design Invited
PDF
Fernando A. Gonzales-Zubiate 1*
1School of Biological Sciences and Engineering, Yachay Tech University, Ecuador. Member of the Scientific Board, Bionatura Journal (role declared for transparency only)
*Corresponding author fgonzales@yachaytech.edu.ec
              
    
ABSTRACT
The CRISPR/Cas9 technology is a game-changer in the field of molecular genetics, enabling the first rapid and efficient genomic editing in virtually any organism. The gRNA is a crucial factor in the technology that determines editing efficiency and specificity. The field has progressively moved away from empirical recipes towards computer-aided precision-engineering methodologies based on artificial intelligence (AI) and deep learning over the last decade. This review discusses important breakthroughs in gRNA design, remaining limitations, and prospects, with an emphasis on translational and biotechnological applications.
 
Keywords. CRISPR, gRNA Design, Genome Editing, Artificial Intelligence, Deep Learning, Machine Learning, Computational Biology, Bioinformatics, CRISPR-Cas9.
 
    
INTRODUCTION
    
CRISPR/Cas9 is now an indispensable tool in modern genome editing, enabling precise genome editing by allowing a nuclease to be directed to essentially any DNA sequence. Nevertheless, its performance largely relies on the careful screening and optimization of the gRNA. Initial design strategies employed basic sequence-based rules—such as avoiding extreme GC ratios or potential secondary structures—yet these failed to incorporate variables like chromatin environment, off-target effects, or species-specific genomic traits 2,3. Recently, the incorporation of computational modeling and multi-omics datasets has revolutionized gRNA design methodologies 1,3,4.

 
 
Key Developments
 
Machine learning/deep learning (ML/DL) is a promising field that has made significant contributions to gRNA efficacy and specificity prediction. Deep learning architectures, such as CNN and RNN, can currently capture complex sequence-structure relationships. Tools like DeepCRISP predict in greater detail, incorporating epigenomic data—such as histone modification, DNA methylation state, and chromatin openness—for context-aware designs.
 
Meanwhile, new CRISPR effectors have added diversity to the editing toolkits. Ultra-small nucleases, such as CasΦ and CasMINI, permit delivery with small size constraints and vectors. Reversible epigenetic editors, like CRISPRoff/CRISPRon, enable transient regulation of genes without inducing permanent DNA cleavage. These advances are expanding the field of CRISPR beyond precision medicine, diagnostics, and agricultural biotechnology 2,6.

 
 
Persisting Challenges
 
 
Despite great progress, there are still several challenges:
 
     
  • Generalizability: Prediction      models work well for human and mouse genomes, but often perform poorly in      AT-rich or repetitive genomic regions in      species other than model organisms.
    •  
  • Model      Interpretability: The “black-box” nature of many AI-driven predictions impedes      their adoption in clinical and regulatory settings, where mechanistic      understanding is essential 3.
    •  
  • Delivery      Limitations: Constraints related to viral vector capacity, immunogenic      responses, and the suboptimal efficiency of non-viral delivery systems      continue to hinder therapeutic applications.
    •  
  • PAM Requirements: Although      engineered Cas variants, such as SpRY, have relaxed targeting constraints,      complete independence from protospacer adjacent motifs (PAMs) remains      elusive.

 
 
Future Directions
 
Progress in gRNA design and CRISPR technology will hinge on:
 
     
  1. Implementing      generative AI and transformer-based models to decode complex      sequence-structure-functional relationships and enhance off-target      prediction 7.
    2.  
  2. Broadening      training datasets to include greater genomic and epigenomic diversity      across species 1,8.
    3.  
  3. Creating adaptive      design pipelines that iteratively refine gRNAs using real-time      experimental feedback 1.
    4.  
  4. Advancing delivery      mechanisms—such as lipid nanoparticles, extracellular vesicles, and      physical methods—to achieve efficient and safe in vivo editing 2.
    5.  
  5. Accelerating      protein engineering efforts to develop nucleases with expanded targeting      ranges and improved fidelity 5.
 
 
 
                   
 
Figure 1. Evolution of CRISPR gRNA design: from empirical rules to AI-driven precision. The field has progressed from simple sequence-based criteria (GC content, secondary structures, repeats) to heuristic algorithms, deep learning–based predictive models, and finally generative AI and next-generation editors (CasΦ, CasMINI, CRISPRoff/CRISPRon), enabling adaptive, context-aware, and clinically relevant gRNA design.
 
 
                               
CONCLUSIONS
From Rules of Thumb to Precision AI-driven: CRISPR-Cas gRNA Design. The evolution of gRNA design methods reflects a clear transition from static, rule-based regimentations to dynamic, context-dependent approaches directed by AI. The combination of computational modelling, synthetic biology, and genome engineering is bringing new levels of control. As these systems become more amenable, interpretable, and robust, they should usher in a new age of personalized therapeutics and biotechnological breakthroughs—altering, once and for all, not just how we edit genomes but also how we probe and engineer the very structure of life.
 
 
Funding
This research received no external funding.
 
Acknowledgments
The authors would like to thank the BioNatura Journal editorial board for their support. No specific grants from public, commercial, or not-for-profit funding agencies were used for this work.
 
Conflicts of Interest
The authors declare no conflict of interest.
 

REFERENCES

1.     Chuai G, Ma H, Yan J, et al. DeepCRISPR: optimized CRISPR guide RNA design by deep learning. Genome Biol. 2018;19(1). https://doi.org/10.1186/s13059-018-1459-4
 
 
2.     Mohr SE, Hu Y, Ewen-Campen B, et al. CRISPR guide RNA design for research applications. FEBS J. 2016;283(17):3232–3238. https://doi.org/10.1111/febs.13777
 
 
3. Hoberecht L, Perampalam P, Lun A, Fortin J. A comprehensive Bioconductor ecosystem for the design of CRISPR guide RNAs. Nat Commun. 2022;13(1). https://doi.org/10.1038/s41467-022-34320-7
 
 
4.      Motoche-Monar C, Ordoñez JE, Chang O, & Gonzales-Zubiate FA (2023). gRNA Design: How Its Evolution Impacted on CRISPR/Cas9 Systems Refinement. Biomolecules, 13(12), 1698. https://doi.org/10.3390/biom13121698
 
 
5. Pausch P, Al-Shayeb B, Banfield JF, Doudna JA. CRISPR–CasΦ from huge phages is a hypercompact genome editor. Science. 2022;369(6511):333–337. https://doi.org/10.1126/science.abf0858
 
 
6. Nuñez JK, Chen J, Pommier GC, et al. Genome-wide programmable transcriptional memory by CRISPR-based epigenome editing. Cell. 2021;184(9):2503–2519.e17. https://doi.org/10.1016/j.cell.2021.03.025
 
 
7.     Nijkamp JF, Woldringh G, Mahfouz A. Foundation models for DNA and protein engineering. Nat Biotechnol. 2024;42(5):567–572. https://doi.org/10.1038/s41587-024-01945-3
 
 
8.     Manguy J, Glover N, Calo E, Jaganathan V. Kingdom-Wide CRISPR Guide Design with ALLEGRO. bioRxiv. 2025. https://doi.org/10.1101/2025.02.13.638206
 
 
Received: June 14, 2025 / Accepted: August 23, 2025 / Published: September 15, 2025
 
Citation: Gonzales-Zubiate FA. From Hack and Tip to AI Precision: the Evolution of CRISPR gRNA Design. Bionatura Journal 2025;2(3):15. doi: 10.70099/BJ/2025.02.03.15
 
Información adicional
La correspondencia debe dirigirse a: fgonzales@yachaytech.edu.ec
 
Información sobre la revisión por pares. Bionatura agradece a los revisores anónimos su contribución a la revisión por pares de este trabajo utilizando https://reviewerlocator.webofscience.com/.
 
ISSN.3020-7886
 
Todos los artículos publicados por Bionatura Journal son de acceso libre y permanente en línea inmediatamente después de su publicación, sin gastos de suscripción ni barreras de registro.
 
Nota del editor: Bionatura Journal se mantiene neutral en cuanto a las reclamaciones jurisdiccionales en los mapas publicados y las afiliaciones institucionales.
 
1.        Copyright: © 2025 por los autores. Se presentaron para su posible publicación en acceso abierto bajo los términos y condiciones de la licencia Creative Commons Attribution (CC BY) (https://creativecommons.org/licenses/by/4.0/).
 
Back to content