Peptides are the short protein chains doing much of your cells' talking
String a few amino acids together and you get a peptide; keep going past a certain size and it becomes a protein. These short chains act as hormones, antibiotics and venoms, and by some estimates they broker up to 40 percent of the protein partnerships inside human cells. At least a tenth of the drug market is built on them.
Chemically, a peptide is a chain of amino acids joined by peptide bonds, usually linear, with a free amine group at one end and a carboxyl group at the other. Under twenty amino acids counts as an oligopeptide; a long unbranched chain is a polypeptide, and once it reaches around 10,000 daltons it is called a protein. Some peptides are rings, and a few even form lasso-like loops. After they are built, cells often decorate them with phosphate, sugar or sulphur groups, and platypus venom goes further, flipping some amino acids into their mirror-image form.
Most are made on the ribosome and then trimmed into hormones and signalling molecules. Others skip the ribosome entirely. Bacteria, plants and fungi assemble many peptides on modular enzyme factories called nonribosomal peptide synthetases, which can bend chains into intricate rings and are closely related to the machinery that makes fatty acids, so hybrid molecules are common. The antioxidant glutathione, found in most oxygen-breathing organisms, is built this way. Some microbes use peptides as weapons against rivals.
Peptides also turn up as practical tools. Peptones, made by digesting milk or meat, feed bacteria and fungi grown in laboratories. Fragments of proteins let scientists identify the original protein, whether from a controlled experiment or a sample degraded by time in a forensic or fossil setting.
Designing new ones has become a computing problem. Because lab screening is slow and expensive, researchers train machine-learning models on sequence and structure data to predict antimicrobial or cell-penetrating activity. They map a chemical space using properties such as weight and fat solubility, where tweaks like cyclisation or unnatural amino acids shift a molecule's stability and target. Biased datasets and hard-to-interpret models remain obstacles.
Source: Peptide
