TIL in a study published in Nature Communications, researchers used a deep-learning system to sift through a database of more than 40 million snake venom peptides. In a matter of hours, the algorithm flagged 386 compounds with the molecular hallmarks of next-generation antibiotics.