Evo 2 is an advanced artificial intelligence model, developed by the researchers of the ARC Institute, Stanford University and Nvidia: trained on 9.3 thousand billion of DNA bases by a highly treated genomic atlas that embraces all the domains of life, this model is able to predict genetic variations and generate genomic sequences in all the domains of life.
Such as Chatgpt, the famous artificial intelligence chatbot (IA) recently available also on WhatsApp, Evo 2 is ready to land In workshops around the world, as an open system (including model parameters, the learning code) to accelerate the exploration and design of biological complexity. Developed by the researchers of the ARC Institute, Stanford University and Nvidia, Evo 2 is an advanced artificial intelligence model able to predict genetic variations and generate genomic sequences in all domains of life.
Tests show that Evo 2 carefully predict the functional effects mutations in procarial and eukaryotic genomes and can also rewrite the wooden mammoth genome From raw genomic sequences, without a direct learning reference, demonstrating the ability to generalize the function from the sequence alone. In the study “Genome modeling and design across all domainins of life with evo 2“Available in preprint on BIORXIVits developers explain in detail how this artificial intelligence model trained out of 9.3 thousand billion of DNA base couplesallows forecasts and design on a genomic scale. Evo 2, the teams specifies, can analyze and generate up to 1 million nucleotides at a time, managing to evaluate long -haul models and relationships within DNA sequences.
During training, Evo 2 has shown that he can predict the next pair of a sequence of a sequencesimilarly to how linguistic models provide for the next word in a sentence. “This approach allows Evo 2 to identify complex genomic structures and to keep the functional impact carefully of genetic variations in all domains of life“ They said the developers.
Now you can use chatgpt on free WhatsApp in advanced form: how to send photos and audio
The test results have in fact highlighted the ability of Evo 2 to predict carefully The functional effects of mutations in the genomes of procarial and eukaryotes, without the need to specify the activity of the genes. “The model has shown sensitivity to mutations in the beginning of the beginning, in the splicing sites and in the preserved genomic regions, with performance in line with the known biological constraints“.
Regarding the generation of sequence on a genomic scale, Evo 2 has proven to be able to Create complete mitochondrial genomesbacterial genomes and sequences on a chromosomal yeast scale. “We decided to make Evo 2 completely open, including model parameters, training code, inference code and opengenome 2 dataset, to accelerate the exploration and design of biological complexity – added the developers -. It represents a significant progress in genomic artificial intelligence, combining predictive accuracy with generative capacities on a genomic scale “.
