A team at the University of Alberta has published the human urine metabolome, a list of more than 3,000 chemicals detectable in the urine of an average, healthy person. The database will help everyone from environmental toxicologists to doctors looking for new diagnostic tests.

U of A biology professor David Wishart and his team trawled through tens of thousands of journal articles using sophisticated computer search algorithms. Once a substance was identified, they then had to purchase or synthesize it in order to provide a known standard against which to measure real human samples (from willing volunteers, including some of the authors of the study). They used gas and liquid chromatography, mass spectrometry, nuclear magnetic resonance imaging and, in the case of compounds at ultra-low concentrations, fluorescent labelling to make them easier to find. “In many cases, the compounds just weren’t there,” says Wishart. “People had misidentified, mislabelled or simply hadn’t been careful enough.” The number of dead ends is partly why the project took so long — more than seven years — but the final list was so thoroughly cross-checked that it was recently published in PLoS ONE without any revisions.

Wishart wasn’t surprised at the overall number; a similar project he led two years ago found more than 4,000 compounds in human blood. Although there were some unusual materials — including platinum and neodymium from environmental exposure — the real power of the database is in allowing researchers to go beyond their current diagnostic tests, which look for a single compound and are therefore rather crude. “If you’re trying to find a single marker for a disease, you’ll fail most of the time,” says Wishart. “But looking for combinations of chemicals that are either too high or too low allows you to come up with a signature.” Wishart says that the literature has proposed chemical signatures for pneumonia, tuberculosis, early-stage colon cancer and even the common cold. The new database provides, for the first time, a benchmark against which to calibrate these tests.