Catch it before it hits the market: DAMD's predictive database teaches devices to recognize designer drugs

"Designer" psychoactive substances are legions of molecules that mimic the effects of known drugs but escape control: synthetics change one fragment in the structure - and standard searches in mass spectral libraries are silent. At the same time, the new formulas are unpredictable in the body and are involved in fatal poisonings. A team of researchers presented the DAMD ( Drugs of Abuse Metabolite Database ) at the ACS Fall 2025 conference - a predicted library of chemical structures and mass spectra of potential metabolites of designer drugs. The idea is simple: if you have "theoretical fingerprints" of future substances and their decay products in advance, the chances of recognizing them in a patient's urine or in a forensic examination increase dramatically.

Background of the study

The market for “designer” psychoactive substances is changing faster than standard laboratory libraries can be updated. Manufacturers deliberately make tiny changes to the structure of known molecules (fentanyls, cathinones, synthetic cannabinoids, new benzodiazepines, nitazenes) to bypass controls and tests. For clinics, this means patients with severe poisonings in whom standard screenings find nothing; for forensic toxicology, it means delayed recognition of “new” substances and the risk of missing substances responsible for fatal cases.

The technical problem is twofold. First, immunoassays are tailored for several “old” classes and are poorly transferred to new analogues. Second, mass-spectrometry panels work like “Shazam for chemistry”: the device compares the spectrum of an unknown peak with a reference in the library. But fresh designer molecules simply do not have such a reference. The situation is complicated by biology: metabolites are more often found in blood and urine, rather than the “parent” molecule. They arise after phase I (oxidation, reduction, hydrolysis) and phase II (glucuronidation, sulfation) reactions, and a whole scattering of derivatives can exist for one original substance. If the library “knows” only the original, the analysis easily misses.

Hence the interest in high-resolution mass spectrometry (HRMS) and in silico tools that predict in advance which metabolites are likely and how they will fragment in a mass spectrometer. Such approaches fill the gap between rare, labor-intensive measurements of reference spectra and the daily need for rapid answers in clinics. The idea is simple: if a lab has theoretical fingerprints of potential metabolites at hand, the chances of recognizing a new substance before it gets into classic reference books increase dramatically.

Organizationally, this is important not only for science, but also for practice. Early recognition of an unknown class allows for faster selection of therapy (for example, immediately thinking about naloxone for opioid intoxication), launching sanitary warnings and adjusting the work of harm reduction services. For forensics, this is a way to work proactively, rather than catching up with the market. However, any "predictive" databases require careful validation: predicted structures and spectra are hypotheses that need to be confirmed by real data, otherwise the risk of false matches increases. Therefore, the current focus is to stitch predictive libraries with already recognized references (like SWGDRUG, NIST) and show added value in real sample flows.

How They Did It: From a “Baseline” Library to Predictions

The starting point was the SWGDRUG (DEA working group) reference database, which contains verified mass spectra of >2,000 substances seized from law enforcement. The team then modeled the biotransformations of these molecules and generated almost 20,000 candidates - putative metabolites - along with their "theoretical" spectra. These spectra are now being validated on sets of "real" data from non-targeted urine analysis: if there are close matches in the array, it means the algorithms are moving in the right chemical space. In the future, DAMD may become a public addition to current forensic libraries.

What's inside the database and how it differs from conventional libraries

Unlike commercial and departmental libraries (for example, the annually updated Mass Spectra of Designer Drugs set), which contain measured spectra of already known substances, DAMD is a forward-looking forecast: digitized hypotheses about what metabolites will appear in as yet unstudied designer molecules and how they will be fragmented in a mass spectrometer. Such "anticipatory" replenishment closes the main gap: the analyst is looking not only for the molecule itself, but also for its traces after metabolism, that is, what is actually found in biosamples.

How it works in practice

Express screening in toxicology works like this: the device receives the mass spectrum of an unknown peak and compares it with a catalog of reference spectra - like Shazam for chemistry. The problem with designer substances is that there is no standard: the molecule is new, the metabolites are new - the catalog is silent. DAMD feeds the device plausible "phantom" standards - spectra obtained by computational modeling for predicted metabolites. According to the team, the set is based on SWGDRUG, replenished with tens of thousands of theoretical spectra and is already run through real catalogs of urine tests. The next step is to demonstrate the proof of principle in forensic toxicology.

Why do the clinic, laboratories and police need this?

• In the emergency room, the doctor sees "suspicious" metabolites in the urine report that resemble fentanyl derivatives - this quickly leads to the correct rescue tactics, even if the original substance was masked in the mixture.
• In forensic toxicology: it is possible to detect “new products” on the market earlier and update methods proactively, rather than reactively - when poisonings have already occurred.
• In resource labs: DAMD can potentially be used as an add-on to existing libraries (NIST, SWGDRUG, commercial assemblies), saving weeks of manual spectrum decoding.

Key facts and figures

• Title and purpose: Drugs of Abuse Metabolite Database (DAMD) - predicted metabolic signatures and mass spectra for "new psychoactive substances" (NPS).
• Where we started: SWGDRUG base with spectra of >2000 confiscated substances.
• Prediction scale: ≈20,000 putative metabolites with "spectral fingerprints"; third-party reviews note a total volume of tens of thousands of theoretical MS/MS spectra.
• Where presented: ACS Fall 2025 paper (Washington, August 17-21), sponsored by NIST.

Technical notes

• Source of "references": SWGDRUG - electron ionization (EI-MS) libraries for seized substances; DAMD - predicted MS/MS metabolites for biospecimens. This is logical: in urine, the decay is more often visible, not the "parent".
• Fragmentation modeling: Press reviews point to the use of high-fidelity CFM-ID simulations to generate theoretical spectra at different collision energies (which increases the chance of agreement across methods).
• Validation: comparison with untargeted urine analysis arrays (lists of all detected peaks/spectra) to filter out unrealistic structures and fit models.

What this does not mean

• Not a "magic wand". DAMD is still a research library, shown at a scientific meeting; it will be introduced into practice after validations and releases for device ecosystems.
• Errors are possible. Predicted spectra are models, not measurements; their reliability depends on chemically plausible metabolic pathways and a correct fragmentation engine.
• The market is flexible. Synthetic producers change their recipes quickly; DAMD wins precisely because it scales and can quickly acquire new predictions, but the race will remain a race.

What's next?

• Pilot in toxicology: show that addition of DAMD to current libraries improves sensitivity and precision for NPS in real-world sample streams.
• Integration with commercial kits: “gluing” with annual releases of designer drug libraries and automatic non-targeted search.
• Transparent release: Make DAMD available to the community (versions, format, metadata) so that it can be used not only by federal labs but also by regional LVCs.

News source: American Chemical Society press release about the ACS Fall 2025 talk, " Building a better database to detect designer drugs "; description of the DAMD project and its validation; SWGDRUG source databases; context on existing commercial libraries.