Like most websites The Translational Scientist uses cookies. In order to deliver a personalized, responsive service and to improve the site, we remember and store information about how you use it. Learn more.
Research Field Biomedical engineering, Microbiology

Taking the Guesswork out of Bacteriotherapy

An imbalance in the microbiome – the micro-organisms living on and in the body – can have serious health implications. Reintroducing “healthy” bacteria (probiotics) to bring the microbiome back onto an even keel has great potential, but it can be hard to predict how a patient’s individual microbiota will react. Now, researchers at Brigham and Women’s Hospital at Harvard Medical School and the University of Massachusetts have created a new tool to help develop better bacteriotherapies. The Microbial Dynamical Systems INference Engine (MDSINE – pronounced M-Design) is a suite of computer algorithms that can forecast microbial behavior in complex ecosystems like the mammalian gut (1).

Gut reaction

The open-source software uses time-series data (a series of data points collected over time) to predict how different interventions will affect the microbial community in the gut – and could help with the development of better probiotics, says Georg Gerber, lead researcher and Assistant Professor of Pathology at Harvard Medical School. “We created MDSINE as a tool to aid research into therapeutic avenues for diseases relating to the gut microbiota, such as inflammatory bowel disease and Clostridium difficile infection.” 

MDSINE can predict how the microbial community will grow and interact, the stable states it will form, and the species most critical to maintaining stability or most vulnerable to perturbations in the system. The basic concept of MDSINE – looking at a system over time and predicting future patterns – is not new. But when dealing with a system as complex as the gut microbiome, the data pose unique problems. “In trying to describe the interactions that happen within these ecosystems, we found that even the simplest models need to be fairly complicated to capture even the key factors,” says Gerber. “We also ran into the problem of too much noise in the datasets, or data that weren’t regularly sampled.” Consequently, the team spent a lot of time developing models that could adjust for the inherent features of microbiome data.

In a recent paper in Genome Biology (1), the group explain how they validated the software with extensive computer simulations, and two real-life studies with mice. In the mouse experiments, they used MDSINE to identify the combinations of microbiota that most effectively inhibit C. difficile infection, and analyzed the impact of diet on a probiotic cocktail intended to treat inflammatory bowel disease (IBD). Both studies gave intriguing results. A combination of three bacteria was predicted to be sufficient to inhibit C. difficile – one inhibits the pathogen directly, while the other two play a supporting role. The second study tested whether the thirteen bacteria in a probiotic cocktail for IBD were affected by changes in the diet of the recipient – an important factor for human IBD patients, who often try dietary interventions alongside medication. The data confirmed that diet does affect the probiotic bacteria, suggesting a new avenue for research.

Forward thinking

Gerber has worn many hats over the course of his career, studying pure mathematics, computer science, computational biology, microbiology, and pathology. He even spent a few years in the digital entertainment industry, before gaining his PhD and MD, and training as a clinical pathologist. Microbiomics was a perfect opportunity to bring his interests together, recounts Gerber; “I saw new technologies like high-throughput sequencing emerging and vastly expanding our ability to understand the microbiome, and I was hooked.” Gerber is now Co-director of the Massachusetts Host–Microbiome Center at Brigham and Women’s Hospital, as well as an assistant professor of pathology at Harvard Medical School.

Gerber and his colleagues made the choice to release MDSINE as open source software for two key reasons, he says. “Firstly, scientific integrity – I think it’s crucial for people in the scientific community to know exactly how we built our algorithms. If your peers can’t look at the source code, they can’t critically evaluate the algorithms. Secondly, we created the software to help make more and better probiotics. I strongly believe that tools meant for academic research shouldn’t be locked up as proprietary code, but should be openly available for the scientific community to use.”

MDSINE is now freely available to researchers, both as source code ( and as an executable open source package (, but the team’s work isn’t over. “We’re going in two different directions with it at the moment; applying the existing algorithms to more scenarios and data sets, and extending the algorithm,” says Gerber. “We named the tool MDSINE for a reason; our focus is on designing probiotics in a rational way. We’re looking at a range of different systems and different questions relevant to that goal.”

The aim is to be able to move on to human microbiome data, though that will require some further work to extend the algorithms. Their published paper used data from mice whose microbiomes consisted entirely of known organisms. These “gnotobiotic” mice provided a controlled system; “Using gnotobiotic mice was great for our proof of principle, but scaling the algorithm for human datasets is definitely a challenge,” says Gerber.

Another challenge is finding human datasets to work with, as Gerber explains, “A lot of time-series studies looking at dynamics of the microbiome are very limited for human data, with only two or three time points per person recorded. For an algorithm like MDSINE we need a minimum of about ten time points. Ideally, the subjects in the study have to undergo some type of experimental perturbation of their microbiome, like a change of diet or a course of antibiotics.” These studies are hard to find, but more are being published every year, and Gerber hopes to be able to apply MDSINE to human data in the near future.

Therapy by MDSINE

Gerber’s lab is also starting to consider applications for MDSINE outside of the gut. “Knowledge of the skin microbiome is increasing, so that is of interest to us. We would expect the skin microbiome to have a much stronger link with the external environment than the gut, which is a more contained ecosystem. So we have been thinking about how we could incorporate these environmental factors,” he says. 

Further afield, the researchers are working on a DARPA-funded project, which involves engineering consortia of interacting microbes. Gerber says the project has allowed them to go from simply studying microbial interactions to directly influencing them: “We’re using MDSINE to learn what the wild-type interactions are, and asking if there are leverage points that we can apply to the genetic circuitry to change those interactions, or even to change the large-scale properties of the ecosystem, like stability. Ultimately, if we can engineer an ecosystem of microbes to adhere to certain behaviors, this really opens up exciting possibilities for therapeutics.”

A possible application for these engineered communities is suggested by the emerging field of fecal transplantation. “As a researcher, I’m very excited about the potential of fecal transplants, but as a physician I believe we need to be cautious,” explains Gerber. “The analogy that comes to mind is blood transfusion. It’s saved many lives, but we’ve also had transmission of infectious agents like HIV and hepatitis. It worries me that there aren’t really any uniform standards for fecal microbiota transplants. I’m hopeful that MDSINE can help us rationally design precisely defined cocktails of microbes for these transplants. MDSINE could also be used to predict who the best donor would be, or even modulate the patient’s pre-existing microbiota to make them more responsive to therapy.”

As well as the ongoing work in his own lab, Gerber is looking forward to seeing how other researchers will apply the software, and is excited to see what datasets researchers will come up with, now they have a robust tool for analyzing their data.

Receive content, products, events as well as relevant industry updates from The Translational Scientist and its sponsors.

When you click “Subscribe” we will email you a link, which you must click to verify the email address above and activate your subscription. If you do not receive this email, please contact us at [email protected].
If you wish to unsubscribe, you can update your preferences at any point.

  1. V Bucci et al., “MDSINE: microbial dynamical systems inference engine for microbiome time-series analyses”, Genome Biol, 17, 121 (2016). PMID: 27259475.
About the Author
William Aryitey

My fascination with science, gaming, and writing led to my studying biology at university, while simultaneously working as an online games journalist. After university, I travelled across Europe, working on a novel and developing a game, before finding my way to Texere. As Associate Editor, I’m evolving my loves of science and writing, while continuing to pursue my passion for gaming and creative writing in a personal capacity.

Register to The Translational Scientist

Register to access our FREE online portfolio, request the magazine in print and manage your preferences.

You will benefit from:

  • Unlimited access to ALL articles
  • News, interviews & opinions from leading industry experts