Generative Biology and Evo 2: Re-Writing the Code of Life in 2040

Evo 2 is not just any old AI. It’s based on transformer architecture, just like the backend of the large language models such as Chat GPT. However, instead of reading and writing human languages, Evo 2 is designed to read and write DNA—the language of life.

The evolution from AI that was first applied to genomic data was not a trivial undertaking. Early AI could not digest the particularity of a long piece of DNA or recognize that minor single-base changes can produce massive differences biologically. To overcome these complications, research and development from teams of researchers followed a novel architecture called Striped Hyena 2. Evo 2 is now capable of processing 1 million DNA bases simultaneously, making it capable of analyzing genomes as large as a microbe.With 9.3 trillion DNA letters from more than 128,000 species, Evo 2 is the most comprehensive AI modeled systems in the world. It can analyze coding and non-coding regions of DNA, identify pathogenic variants associated with disease, and it can even create entirely new CRISPR systems—many of which have been synthesized, tested and shown to work in real labs. 

Genetic Mysteries and the Predictive Power of Evo 2

There are many mysteries in genetics, even with mapping the human genome. In fact, approximately 0.1% of your genome is different from the “reference human genome.” That difference could change your life.

Some differences are benign—think of the difference of eye color. Some differences can be deadly e.g., sickle cell anemia, breast cancer; however, many differences fall into a variant of unknown significance (VUS)—a genetic change with indeterminate effects.

Here is where Evo 2 sheds light on mysteries. Traditional biology cannot tell us what the difference is when the variation falls in what is 98% of the genome, which is considered to be junk DNA. Evo 2 can analyze non coding regions of the genome with a high degree of accuracy. In a case about the BRCA1 gene, linked to breast cancer, Evo 2 predicted the pathogenic potential of unknown variants with 90% accuracy, making it outperform all other AI modeled systems.

Importantly, this is not just theoretical, but it can have real, tangible utilities.Patients who had previously received ambiguous genetic test results now have actionable information, and clinicians are able to make informed decisions regarding preventative and therapeutic pathways.

Polygenic Disorders Present a Unique Challenge for AIs.

Unlike single-gene diseases (such as cystic fibrosis, fragile x syndrome, etc.), many disorders that are classified as “diseases” actually result from the interaction of thousands of variants (such as diabetic, heart disease, and schizophrenia). Such complex interactions are tough for humans to spot, but for an AI with trillions of information points to cross-reference, it is manageable.

Evo 2 can analyze genetic data from millions of individuals, and through its identification of underlying correlations, point to the combination of variants that are most likely to result in a condition. This places the clinician in a position to make an accurate diagnosis (or to at least see where the potential problems may lie) much earlier than would have been previously possible and makes personalized medicine available at scale.

From Lab to Clinic: Taking Advantage of Concept Generation with Evo 2

What Evo 2 can also do, besides analyzing, is produce things. If you needed therapy for a rare genetic disease, rather than waiting a substantial amount of time for pharmaceutical development, Evo 2 would be able to create a therapeutic protocol for your condition within days, be it editing faulty genes, designing synthetic proteins, or producing a novel biological system.

To be clear, by 2040, generative biology has empowered AIs to produce custom CRISPR therapies, individualized vaccines, and genetically customized regenerative therapies. AI designed proteins can be printed using bioprinters to aid rebuilding damaged tissues or even organs.

This revolution in the healthtech space will go well beyond humans.Evo 2 has been used to develop crops capable of surviving droughts, bacteria for bioremediation projects that confront polluted environments, and even synthetic life capable of producing clean sustainable fuel.