At a crime scene, a stain is found. In the past, an expert would drip a special solution onto it, and a strip would change color if the stain contained blood. But if blood was mixed with saliva, the strip could be fooled. Now there is another approach: a device called a mass spectrometer weighs protein fragments left in the stain. Each fluid leaves its own set of such fragments, like a signature.
Scientists have come up with three ways to read this signature. The first method is like a dictionary: it lists 526 protein fragments that occur only in one of the five fluids. If the device finds such a fragment, it immediately identifies the fluid. The second method compares how many of one fragment there are relative to others. For example, blood has a lot of one protein, while saliva has almost none. This balance can reveal what was in a mixture. The third method was not taught in advance: it was given many examples and found patterns on its own. This method turned out to be the most accurate: on pure stains it never erred, and on mixtures it erred less than once in twenty cases.
The most interesting part is that all three methods use largely different protein fragments: only 15 out of hundreds were shared. This means they complement each other, like three witnesses who saw the same event from different angles. Today, this approach can already be used in forensics, especially when you need to distinguish semen from skin cells. So far, it has only been tested on pure samples and laboratory-made mixtures. The next step is to test it on real stains from crime scenes, where proteins may have degraded due to time or heat.
Among the 526 specific peptides are fragments of proteins already used in routine immunochromatographic tests: hemoglobin, α-amylase, prostate-specific antigen, semenogelin, and uromodulin. But the proteomic approach expands the panel to hundreds of peptides, greatly increasing specificity.
In the TV series 'CSI: Crime Scene Investigation', experts often determine the type of biological fluid with a single quick test. This work brings reality closer to fiction, showing that mass spectrometry and machine learning can provide an equally fast and far more reliable answer, even for mixtures.