From Nature-Limited To AI-Designed: The Future Of Gene Editing

By Amy Pooler, Chief Scientific Officer and Matt Nethery, Principal Data Scientist, Computational Biology

Naturally occurring gene editing enzymes, shaped by evolutionary pressures rather than clinical objectives, often present significant barriers to therapeutic development. Unengineered variants frequently suffer from limited targeting windows, off-target modifications, and unwanted bystander edits at adjacent nucleotide positions. Shifting from traditional discovery to generative artificial intelligence fundamentally redefines protein engineering. By training protein language models on vast genomic and metagenomic datasets, computational platforms can generate novel enzyme architectures distinct from natural evolution. Integrating these machine learning frameworks with empirical closed-loop screening drastically accelerates editor refinement. Iterative optimization cycles boost editing potency exponentially while enforcing stringent target specificity and reducing bystander activity. This purpose-built approach applies across diverse modalities, including deaminases and large serine recombinases, unlocking previously inaccessible genomic loci and expanding the scope of treatable genetic conditions.

Read the full insight to see how generative design is accelerating precision gene editing.