The Future of Biology: AI Revolutionizes Genome Design

 

Artificial intelligence could play an even bigger role in the field of biology in the future. Scientists have programmed an AI that can create new genomes. We'll summarize for you how this works and what effects it could have.Scientists at the Arc Institute, Stanford University and the University of California have created an artificial intelligence that could accelerate biological research in the future. The AI ​​is called Evo.

This is how scientists created a DNA AI

Evo is able to design new genomes independently. The AI ​​is based on the principle of large language models such as ChatGPT, which string together text blocks to provide an answer. Instead of text blocks, Evo puts together individual DNA, RNA and protein blocks to create individual nucleotides or even new genomes. The basis for this is training with 2.7 million genome data sets. These mainly included genomes of bacteria and viruses that can infect bacteria."Evo decodes the patterns written in DNA over billions of years of evolution," says Patrick Hsu, Stanford Assistant Professor of Chemical Engineering. He adds, "Just as generative AI has revolutionized the way we interact with text, audio and video, these creative possibilities can now be applied to the fundamental codes of life."

What Evo can already do

To test the artificial intelligence, the scientists first gave Evo the task of creating a CRISPR system. This allows individual DNA strands to be severed at a specific point in order to insert new DNA building blocks or otherwise modify the strands. The CRISPR systems consist of proteins and RNA.

As the researchers point out, new CRISPR systems are usually only discovered by "searching nature's existing systems." Instead, the AI ​​was able to suggest several new systems. The scientists had a hit with the eleventh system. "EvoCas9-1" only shares about 73 percent of its structure with CRISPR-Cas9 and otherwise differs significantly from other systems. Nevertheless, it can record similar activities in separating DNA strands.

"Creating functional CRISPR systems requires intricate coordination of proteins and RNA. Evo's ability to bring these two components into harmony and make them work efficiently together demonstrates a new level of sophistication in the engineering tools of biology," Hsu emphasizes.

The researchers also used Evo to create a DNA sequence that can move within genomes. A particularly complex group of these jumping genes is called the IS200/IS605 transposon. Despite the challenge, Evo was also able to create a new set of transposons that are different from any others found in nature.

The outlook for Evo's future

Evo is currently able to design genomes with more than a million base pairs. In the future, the researchers want to significantly increase this output and thus better understand multicellular organisms. Hsu says: "In the long term, we are working towards creating a new field of genome design in which we can do basic cellular work and potentially create new organisms."

Another goal is to reduce the hallucinations that Evo still produces and one day eliminate them completely. These false statements, which many artificial intelligences suffer from, manifest themselves in Evo in incorrect CRISPR systems that cannot function.