Microbes are full of undiscovered molecules, including antibiotics. But many species remain difficult to culture or simply contain large swaths of genes that remain inactive under laboratory conditions. Hoping to harvest untapped molecules without growing a single cell, Sean F. Brady of Rockefeller University and colleagues looked to genomic data from the human microbiome for inspiration. “Genomic sequences are coming at us fast, and we need tools to turn that information into molecules,” Brady says. So the researchers mined previously reported DNA sequences for nonribosomal peptide gene clusters. With those data, they synthesized the predicted structures of 25 peptides and tested them for antibiotic activity. Two peptides, dubbed humimycin A and humimycin B, are effective against several pathogenic bacteria. Humimycin A drastically improved survival in mice infected with methicillin-resistant Staphylococcus aureus (MRSA) when given in conjunction with the antibiotic dicloxacillin (Nat. Chem. Biol. 2016, DOI: 10.1038/nchembio.2207). “This was just the first step of building complex molecules guided by bioinformatics,” Brady says. Next, his group plans to explore beyond the human microbiome to other microorganism habitats.