A new digital library based at UC San Diego aims to massively increase the number of drugs that can be easily detected in samples collected from patients, potentially allowing doctors and researchers to delve more deeply not just into what individuals may have taken or have been exposed to, but also into how those substances interact with each other in the body.
A paper published online Tuesday in the scientific journal Nature Communications describes the creation of the Global Natural Product Social Molecular Networking Drug Library, which allows broad analysis of drugs in patients’ systems by testing samples that range from breast milk to blood.
Researchers collaborated with myriad organizations to pull together detailed data on 4,723 unique drugs, including information on nearly 100,000 elements they break down into. This work includes “fingerprints” for each molecule, which scientists often call “spectra,” a specific value produced when compounds are subjected to mass spectrometry, a common analytical tool that determines chemical composition.
Mass spectrometers can estimate all constituent chemicals in a sample, providing not just which molecules are present but also how much of the total volume each element contributes.
Here is where the UCSD-led effort takes things up a notch. While hundreds of different molecules are already routinely analyzed in medical and environmental testing, analysis tends to focus on a narrow set of targets. Untargeted analysis has been less common because there has been no single source of information available to help decode a large percentage of the readings that mass spectrometry analysis returns.
“Usually, we only understand, let’s say, 5% of all of the recorded molecules, and the other 95%, we know that there is a molecule there, but we just don’t know what it is,” said the paper’s co-first author Nina Zhao, a chemist and post-doctoral scientist in the laboratory of UCSD bioanalytical chemist Pieter Dorrestein.
But that is not because no one knows what the readouts and characteristics are for that other 95%. It’s just that no one has pulled the information into one compendium that can be easily referenced by the computer programs that analyze results. The key was to link drugs’ spectrographic fingerprints with the rest of the information about them, adding critical context that, with a single click, provides a full pharmacological readout, including information on what each drug is used to treat, how chemicals work in the body and even known sources of exposure.
Sources fall into five broad categories: medical, endogenous or of a person’s own body, food, personal care and industrial.
The research team included a global cast of collaborators from institutions in the Czech Republic, Norway and Spain. This group has already used the library to analyze samples collected from nearly 2,000 participants in the American Gut Project, an initiative that studies the diversity of microbes in the human digestive tract. And tests of more than 3,000 food products proved able to spot everything from antibiotics in meat to pesticides in vegetables.
“This library is really helping us make connections between drugs and different health conditions and diseases, so it’s really helping us to discover new things,” said co-first author Kine Eide Kvitne, a post-doctoral researcher and pharmacist in the lab of co-author Shirley Tsunoda.
Analysis of samples has generally shown what researchers expected; for example, those with inflammatory bowel disease showed a high frequency of antibiotics and skin swabs from subjects with psoriasis showed strong signals for antifungal agents often prescribed to those with skin lesions.
But there were also some differences visible across regions and genders. Americans were found to carry more detectable drugs than participating Europeans and Australians. Painkillers were more common in women; erectile dysfunction drugs were found in men.
While spotting what’s expected helps to validate methods, there were also indications that being able to identify a broad range of drugs can help researchers dig deeper. Investigators were able to group HIV-positive patients based on the drugs they were taking and observe differences in how their bodies processed those medications.
“We noticed that, despite them all being infected, their different drug usage was connected to changes of their gut microbiome,” Zhao said.
The ability to identify and annotate a larger group of drugs, then, expands the possibilities of metabolic analysis, potentially spotting interactions that were not visible with more-targeted analyses.
Michael Snyder, a professor of genetics at Stanford University and author of the book Genomics and Personalized Medicine, said that having such a broad library does offer the potential to identify less commonly targeted molecules, whether they come from drugs a person takes or from their environment.
“If you want a holistic picture of what’s going on in a person, you do need that information, so I do think this could give a more complete picture,” Snyder said.
One particularly intriguing area where a broad library could be relevant, he added, is in drug interactions.
“Let’s say you’re taking a bunch of drugs,” Snyder said. “We really don’t know how they interact with each other and, if we had enough of those spectra, we’d probably learn something pretty interesting.” Added spectra to second graf where we discuss fingerprints
And there are also efforts underway to use the library to spot drugs out in the world. In one instance, a ski resort is using the library to detect the possible presence of drugs in snow made from reclaimed water.
