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.<p>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.<p>[0] <a href="https://www.cell.com/cell/fulltext/S0092-8674(24)00522-1?_returnURL=https%3A%2F%2Flinkinghub.elsevier.com%2Fretrieve%2Fpii%2FS0092867424005221%3Fshowall%3Dtrue#%20" rel="nofollow">https://www.cell.com/cell/fulltext/S0092-8674(24)00522-1?_re...</a>
I was expecting novel modes of action but unfortunately it is not the case here:<p>> Our findings indicate that the tested AMPs from AMPSphere primarily exert their effects by permeabilizing the outer membrane rather than depolarizing the cytoplasmic membrane, revealing a similar mechanism of action to that observed for classical AMPs and EPs from the human proteome.
These are all linear peptides they have close to 0 interest for direct applications as antibiotics because your body likes to chew on them.
But they can be used to discover new targets or to design more resilient molecules (cyclic peptides, hybrids...)
It shouldn't be too surprising that these are peptides. It should be easier for a bacterium to evolve a new peptide than would be to evolve a new non-peptide antibiotic, since the latter requires evolving a new sequence of enzymes, each of which is a peptide.
Link to their website: <a href="https://ampsphere.big-data-biology.org/home" rel="nofollow">https://ampsphere.big-data-biology.org/home</a>