Follow
'On the assumption that a human SNP with commonly observed counterparts in primates probably doesn’t cause disease, Farh exonerated many human variants. His team also used the “benign” primate SNPs to train a neural network, called Primate AI-3D. With AlphaFold, a protein-structure prediction tool based on artificial intelligence (AI), as its scaffold, his program builds 3D models of each protein. Based on the benign SNPs, it identifies regions where changes to the protein’s structure would not disrupt its function. Conversely, changes in other regions were more likely to cause problems.'
