The company created a green fluorescent protein using its AI model (representative image).

New Delhi:

A New York-based AI research company that focuses on biotechnology has developed a new protein using protein language models (ProLLMs) that operates on a transformer architecture similar to ChatGPT.

EvolutionaryScale on June 25 unveiled a new AI-generated protein molecule that glows—mimicking the bioluminescence of a jellyfish molecule called green fluorescent protein. The sequence of the novel protein is significantly different from the natural protein (less than 60 percent similarity): a difference the company calls “more than 500 million years of (natural) evolution” possible.

The company used its frontier AI language model called EvolutionaryScale Model-3 (ESM3) to achieve this feat and secured $142 million in a seed funding round, which included the likes of Nvidia and Amazon. Investments of industrial companies are also included.

We are thrilled to partner with AWS and NVIDIA to push the frontier of AI for the life sciences.

— EvolutionaryScale (@EvoscaleAI) June 25, 2024

ESM3 differs from ChatGPT in that it is trained on parameters (intrinsic variables) of three fundamental biological properties of proteins – sequence, structure and function. The model was trained on 98 billion parameters, making it the largest biological AI model to date.

EvolutionaryScale calls it a “model trained in all evolution”. The training set consisted of 2.78 billion natural proteins, including “microbes in extreme environments such as the Amazon rainforest, deep oceans, hydrothermal vents, and a handful of soils.”

ESM3 lets users design proteins, using prompts with partial information (sequence, structure, and function keywords) and repeating the model to make predictions until the complete sequence is complete. This model is primarily for scientists and gives them unprecedented control over the process of making proteins.

We’ve trained ESM3 and are excited to introduce EvolutionaryScale.

ESM3 is a generative programming language model for biology. In experiments, we found that ESM3 can simulate 500M years of evolution to create new fluorescent proteins.

Read more:

— Alex Rives (@alexrives) June 25, 2024

EvolutionaryScale states that their goal is to make organisms programmable. “ESM3 takes a step toward a future where AI is a tool to engineer biology from first principles in the same way we engineer structures, machines and microchips, and write computer programs,” the company’s website explains.

The application of this technology could lead to breakthroughs in a number of fields such as drug discovery and development, biomedical research, as well as sustainability – as already demonstrated by EvolutionaryScale’s demonstration of a protein prototype that is made of plastic. Has the ability to reduce waste.

The possibilities are endless, every cell in every organism contains ribosomes (protein complexes responsible for protein synthesis). However, there have also been concerns that AI could be misused to create biological weapons.

The scientists took a proactive approach and in March set out “social values, guidelines, and commitments for protein design for the responsible development of AI”, to guide progress in this domain for the betterment of humanity. trying to

We are promoting a new global agreement signed by 100+ leading scientists to ensure that AI technologies for protein design are developed responsibly. This field can provide drugs, vaccines and other innovations that benefit everyone.

Institute for Protein Design (@UWproteindesign) March 8, 2024

EvolutionaryScale has also been praised by experts for releasing a small, open-source version for others to use freely. The full model has not been widely released, although its training process has been made public, in an effort to remain transparent and share technology freely.

Source link