Microbiomes, the invisible ecosystems of microorganisms that live within and around us, hold profound implications for health, science, and even crime-solving. In this Q&A, SCINQ interviews Eran Elhaik, a pioneering researcher and creator of the MGPS (Microbial GPS) tool, which uses microbiome data to predict the origins of microbial samples. From exploring the challenges of contamination to discussing its revolutionary applications in forensics and healthcare, Elhaik reveals how this groundbreaking technology is reshaping our understanding of the microbial world. Dive into this conversation to discover how microbiomes are unlocking new frontiers in science and innovation.
Let’s start with the basics. Although microbiomes are widely discussed, many people only associate them with the gut. Can you explain what the microbiome is and its broader implications?
Absolutely! The microbiome refers to the collection of microorganisms—bacteria, viruses, fungi, and others—that live inside and on the human body.
Most people know about the gut microbiome, which includes everything in our digestive tract, like the stomach and intestines. But we also have microbiomes on the skin, in the mouth, and even surrounding us as an aura of bacteria. Each body site has its own unique microbiome.
These microorganisms are incredibly dynamic. They change based on our age, diet, gender, and even our daily activities. Some bacteria are stable over time, while others change rapidly. However, we only know a fraction of them because most cannot be cultured in a lab.
This complexity makes microbiomes a major focus in medicine, especially as we’ve started to understand how restoring microbiome balance could treat diseases. But it’s still a painstaking process because we don’t know the names or functions of most microorganisms.
What inspired you to develop the MGPS tool, and what challenges did you face?
The idea came from my work on the MetaSUB project, which maps microbiomes in subway systems worldwide. Back in 2014, Chris Mason from Cornell launched this international effort, and I joined as the data analysis lead.
At the time, I had already developed tools for human biogeography, like GPS Origins, which uses human DNA to predict where someone’s ancestors came from. I wondered if we could apply a similar approach to microbiomes.
But this was no easy task. For one, microbiome data varied widely depending on how and where it was collected—different tools, technologies, and protocols. We couldn’t combine datasets easily, so we had to analyze each one separately.
Another challenge was the lack of data. Machine learning algorithms rely on large datasets, but global microbiome datasets are often very limited. For example, some datasets had only 130 samples—far too few for traditional machine learning approaches. We had to design a clever algorithm that could work effectively with limited data by focusing on differences in bacterial frequencies across locations.
You mentioned contamination earlier. How does it affect your work?
Contamination is a big issue, especially with microbiomes. With human DNA, you can ask someone about their ancestry to validate the results. But with microbiomes, you don’t always know what you’re collecting.
For example, a tourist from Australia might touch a subway rail in London, leaving behind bacteria from Australia. If you sample that rail, you might think it represents London when it doesn’t. To address this, we take multiple samples from each site and use statistical methods to account for these anomalies.
Does MGPS work better in some environments than others?
Definitely. The algorithm performs best in environments where microbiomes vary significantly between locations. For example, rural areas might lack enough data for detailed predictions, while crowded urban areas like central London pose their own challenges.
In places like central London, you have a high density of people constantly moving around, which “smears” the microbiome and makes it homogeneous across locations. Our algorithm assumes that microbiomes differ from place to place, so this kind of homogeneity reduces its accuracy.
How does MGPS compare to traditional tools in microbiology and forensics?
MGPS is a breakthrough because it predicts microbial origins, not just correlations. Traditional tools can match bacteria to known locations but can’t infer where unseen samples belong.
Think of it this way: traditional tools are like matching shapes to holes in a toy. MGPS, on the other hand, learns the principle behind the matching process. It can predict where a new shape fits, even if it’s never seen that shape before.
Can you walk us through a forensic application of MGPS?
Imagine a criminal wipes their fingerprints from a gun, removing all DNA evidence. They claim they were at a party and have an alibi.
With MGPS, we can swab their hand, the gun, and the crime scene to match microbiomes. If their microbiome matches the gun and the crime scene but not the party, we’ve undermined their alibi. Microbiome evidence is harder to erase than fingerprints or DNA, and it opens a whole new dimension in forensic science.
How long does a microbiome signature last on objects?
It depends. On personal items like phones or toothbrushes, microbiome signatures are very stable because no one else uses them. On other objects, like a gun, the signature starts to decay after about two weeks. Environmental factors like humidity and temperature can also speed up degradation.
Can MGPS address issues like antibiotic resistance?
Absolutely. When we sequence bacteria, we can identify antimicrobial resistance (AMR) genes and trace where they originate.
For example, if AMR genes appear in England, MGPS can determine whether they came from countries with less restrictive antibiotic use. This knowledge helps healthcare providers implement targeted measures, like stricter sanitation protocols at specific entry points.
What were some surprising insights from your work?
One surprising finding was that tourists weren’t the main contributors to spreading AMR genes, as we initially thought. Instead, it might be goods or produce.
We also learned that the layout and activity of cities, like central London’s density, can significantly affect microbiome patterns. Another surprise was the enthusiasm from law enforcement. When I presented MGPS, several agencies said they had datasets but didn’t know how to use them. They were eager to adopt the technology.
How might MGPS evolve with more global microbiome data?
As more cities participate in projects like MetaSUB, MGPS will continue to improve in accuracy and scope. With data from 60 countries, we can already assist in international crime investigations, like tracking illegal wildlife trade.
Eventually, as we collect more data from underrepresented regions, MGPS could map the entire globe, enabling even broader applications.
What are the future prospects for MGPS?
In forensics, the next step is to collaborate with law enforcement to sample cities and create detailed microbiome maps. This will allow us to deploy MGPS in real-world investigations and solve crimes that were previously unsolvable.
In healthcare, MGPS could help trace the origins of diseases or identify AMR hotspots. With enough data and standardization, the possibilities are endless—from combating antibiotic resistance to even identifying the origins of historical artifacts using microbiome analysis.
If you enjoy the content we create and would like to support us, please consider becoming a patron on Patreon! By joining our community, you’ll gain access to exclusive perks such as early access to our latest content, behind-the-scenes updates, and the ability to submit questions and suggest topics for us to cover. Your support will enable us to continue creating high-quality content and reach a wider audience.
Join us on Patreon today and let’s work together to create more amazing content! https://www.patreon.com/ScientificInquirer