4 ms·
From a quick skim of the paper [0], they discovered ~1M antibiotic-like peptides with an ML model, synthesized 100 of them and saw ~60% of them successfully red
by mmastrac 2y ago
From a quick skim of the paper [0], they discovered ~1M antibiotic-like peptides with an ML model, synthesized 100 of them and saw ~60% of them successfully reduce human-antibiotic-resistant bacterial loads in a mouse model by 2-3 orders of magnitude.
IANAScientist, but this seems like a useful result in that they have a model that predicts antibacterial activity reasonably well, and they have a potentially large number of compounds to explore to treat AB resistance.
[0] https://www.cell.com/cell/fulltext/S0092-8674(24)00522-1?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS0092867424005221%3Fshowall%3Dtrue#%20 https://www.cell.com/cell/fulltext/S0092-8674(24)00522-1?_re...
- b800h 2y agoDoesn't using an ML model mean that they're likely to find antibiotics which are similar enough to existing (and natural) antibiotics to render them potentially ineffective?
- Centigonal 2y agoAccording to GP, they synthesized 100 of them (created the actual molecules) and saw that ~60% of them successfully worked on antibiotic-resistant bacteria.