Next-gen model decodes secrets of genes with near-accuracy [Video]

Call it the next frontier in machine learning: scientists have designed a large language model, Evo, to interpret biological sequence data or genetic code.

To pivot AI in the direction of biology, scientists from the Arc Institute created version 1.0 of “a genomic foundational model,” the genome being a sequence encoded with DNA, RNA, and proteins, trained with millions of microbes instead of words.

With this robust amount of biological data, the model works much like language learning models (LLMs) such as Google Gemini, as per Live Science, or the GPT in ChatGPT. Among its capabilities, it can learn and predict how small changes to the sequence will affect the whole organism, making this potentially groundbreaking AI a powerful tool in …