AI + CRISPR: The explosive fusion that will redefine biotechnology

The integration of artificial intelligence (AI) in the biotech industry seems boundless. Its combination with CRISPR could be the one area where AI has the most potential.

Nobel prize winner and co-inventor of the CRISPR technology, Jennifer Doudna, recently highlighted the profound implications of combining AI with CRISPR in an article published by Wired. She emphasized that AI’s capacity to analyze vast genomic datasets accelerates the discovery of more efficient gene-editing tools and enhances the precision of genetic modifications.

So how is AI set to transform CRISPR? Let’s find out! 

The synergy: Why AI and CRISPR work so well together

While the CRISPR technology is revolutionary, the cost is still significant – around $2 million per patient. Also, the way CRISPR is applied requires a bone marrow transplant, which is far from an easy procedure. 

This is where AI can play a considerable role in the future. AI accelerates the identification of novel CRISPR-associated proteins, such as smaller Cas variants, which are crucial for efficient gene editing. For instance, researchers at the Innovative Genomics Institute utilized AI-based structural searches to discover previously unidentified Cas13 enzymes, expanding the CRISPR toolkit for RNA targeting. AI-accelerated discovery doesn’t stop at the identification of novel CRISPR proteins and enzymes; it could also help find more effective delivery methods.

Precision in gene editing is paramount, and AI significantly contributes by predicting and minimizing off-target effects. Machine learning models like DeepCRISPR and CRISPR-M analyze genomic sequences to design optimal guide RNAs (gRNAs) that enhance targeting accuracy. These AI-driven tools reduce unintended edits, thereby increasing the safety and efficacy of CRISPR-based interventions.

Conversely, CRISPR technology provides a dynamic platform for training and refining AI models. The precise and programmable nature of CRISPR-induced genetic changes offers a controlled environment for AI to learn and predict biological outcomes. This symbiotic relationship enables the development of more accurate AI algorithms tailored for complex biological systems, meaning AI’s integration with CRISPR will only become more complete with time. 

The frontier: Who is integrating AI with CRISPR

The convergence of AI and CRISPR gene-editing technology is propelling biotechnology into new frontiers. Several innovative companies are at the forefront of this synergy.

Mammoth Biosciences

Co-founded by Jennifer Doudna, Mammoth Biosciences is pioneering CRISPR-based diagnostics and therapeutics. At the heart of Mammoth’s approach is its AI-driven metagenomic discovery platform.

Metagenomics involves studying the DNA of all microorganisms – viruses, bacteria, archaea – found in environmental samples like soil, water, or even the human microbiome. This produces an enormous volume of genetic data. Mammoth uses AI algorithms to analyze these datasets, searching for novel CRISPR-associated proteins that are smaller, more efficient, and easier to deliver into human cells than the well-known Cas9.

Manually sorting through these massive datasets to identify functional proteins would take years. AI accelerates this process by recognizing patterns in protein structures and predicting their potential function and effectiveness.

These CRISPR tools overcome delivery challenges, particularly for therapies requiring delivery via viral vectors or lipid nanoparticles, making in vivo gene-editing therapies feasible and safer.

In April 2024, Mammoth Biosciences entered into a collaboration with Regeneron Pharmaceuticals to research, develop, and commercialize CRISPR-based gene-editing therapies targeting multiple tissues and cell types. This partnership leverages Mammoth’s ultra-compact CRISPR systems and combines them with Regeneron’s targeted delivery technologies. AI also plays a role in optimizing guide RNAs and reducing off-target effects, enhancing the precision and efficacy of the resulting in vivo gene-editing treatments.

Profluent

Profluent, a California-based startup, uses large language models (LLMs) – the same AI technology behind tools like ChatGPT – to design entirely new CRISPR proteins that do not exist in nature.

LLMs are trained on vast protein datasets, learning the patterns and structures of naturally occurring proteins. Profluent uses these AI models to generate proteins with specific characteristics optimized for gene editing, such as improved accuracy, stability, or smaller size.

Naturally occurring CRISPR systems, like Cas9, evolved in bacteria to fight viruses – not for human gene editing. While they work remarkably well, they have limitations in precision, efficiency, and delivery. Profluent’s AI-designed proteins can overcome these limitations, offering tools that are purpose-built for therapeutic and research applications.

To promote innovation, Profluent has launched its OpenCRISPR initiative, making these AI-designed proteins available through an open-source license. This democratizes access to cutting-edge CRISPR tools, fostering global collaboration in gene editing.

Beam Therapeutics

Beam Therapeutics focuses on a specialized form of CRISPR technology called base editing, which allows precise changes to individual DNA letters without causing double-stranded breaks in the genome.

Base editing requires extreme precision to avoid unintended edits. AI helps design optimal guide RNAs that target the exact location in the genome while minimizing off-target effects. AI also predicts the outcomes of edits, ensuring accuracy in therapeutic applications.

Base editing is ideal for treating diseases caused by single-letter mutations like sickle cell disease. AI enhances the safety and efficiency of these therapies, bringing them closer to clinical reality.

The road ahead: What CRISPR and AI mean for the future

Doudna told Wired she anticipates that by 2025, AI and machine learning will significantly amplify the impact of CRISPR genome editing in medicine, agriculture, and climate change. CRISPR is an amazing tool, and the most evident application of it is in the development of therapies. However, the future of CRISPR is broader – genome editing also has a big part to play in climate change, another challenge mankind faces. Editing plants and crops could make agriculture more resistant to the climate changes we are facing. 

In the interview we mentioned earlier, Doudna gave the example of using CRISPR to make a methane-free cow, which according to her, is perfectly feasible scientifically. It’s already something we are aiming for by applying precise modifications to the cow’s diet, but genome editing could be used to modify the cow’s microbiome directly to achieve this goal. What the future holds is technologies coming together in cross-industry applications. 

In an interview given to the Financial Times about his Nobel Prize, Demis Hassabis, CEO of Google DeepMind, emphasized that AI tools like AlphaFold, which predicts protein structures, are revolutionizing drug discovery and material design, potentially leading to cures for various diseases and advancements in energy solutions. 

CRISPR earned Doudna the Nobel Prize in 2020 – four years later, it’s the AI tool Alpha Fold that distinguishes itself. The combination of the two technologies holds immense promise in healthcare and beyond. 

It took Casgevy 11 years to be approved but according to experts, AI will allow us to discover and advance more candidates faster in the future.