B · Normal
[AI Model Interprets the "Grammar" of Three Billion Bases] On September 11, a team from the University of California, Berkeley, created the genome language model GPN-Star, which can interpret the "grammar" of three billion bases and identify human genetic variations faster and more accurately. The model uses whole-genome alignment data to learn from evolution and is trained at a much lower computational cost than larger models. The team simultaneously released genome-wide predictive annotations for biologists to screen pathogenic variants and regulatory elements for priority experiments. A related paper was published in the journal Nature on the 9th.
Comments