.webp)
Future applications of ESM3 involve the development of drugs, sustainability and antibody engineering
EvolutionaryScale, founded by former meta-scientists, has launched an innovative AI tool, ESM3, designed to revolutionize protein engineering. Securing $142 million in funding, this model, boasting more than 100 billion parameters, is expected to revolutionize biological AI. Comparable to text-based AI outputs, ESM-3, marketed as a "protein language model," uses its training data from 2.7 billion protein sequences to produce novel proteins that are customized to user requirements.
In a key demonstration, ESM-3 successfully regenerated the Green Fluorescent Protein (GFP), yielding new variants with promising fluorescence properties. Notably, one strain, esmGFP, shows structural similarity to its natural counterparts but with a substantially different amino acid sequence, challenging the conventional evolutionary timeline. While praised by experts like Anthony Gitter as revolutionary, concerns arise about the potential overhyping and social consequences of rapid protein synthesis. EvolutionaryScale has been actively negotiating with regulatory bodies, adhering to strict risk mitigation protocols. Future applications of ESM3 involve the development of drugs, sustainability (including plastic-degrading enzymes), and antibody engineering. However, barriers related to accessibility persist due to the computational requirements of the model, which limit replication in academic settings.
The launch of ESM-3 highlights the tremendous advances in AI-powered protein synthesis, which are expected to reshape biotechnology landscapes across the globe.
.webp)
