Antibiotic usage and resistance

Antibiotics have been one of the most influential medical advances in history. Unfortunately, human antibiotic manufacture and usage makes those same antibiotics less effective. If we cannot produce new classes of antibiotics quickly, then our best hope to avoid a post-antibiotic era is effective stewardship of the antibiotics we have. There is evidence that, if antibiotic use is properly managed, resistance can be kept at acceptable levels, but there are large gaps in our understanding of how resistance emerges and declines in response to antibiotic use.

Specifically, I am using medical records and hospital antibiograms to study the relationship between antibiotic consumption and antibiotic resistance. If we reduce prescriptions for a drug by some amount, by what amount could we expect resistance to reduce?

Previous projects

Most of my work during my PhD focused on environmental microbiology, the human microbiome, and clinical trial design.

Salty diets and the gut microbiome

In collaboration with physician-scientists at the Max Delbrück Center for Molecular Medicine in Berlin, I helped investigate the effects of a salty diet on the mammalian microbiome and host.

Microbial petroleum degradation

In collaboration with the Hazen Lab at the University of Tennessee (Knoxville), I developed an analytical method, texmex, to identify important organisms involved in oil degradation. This tool help disentangle the “bottle effect”, the perturbation caused by removing microbes from their natural ocean environment, from the effects of interesting treatments, like added oil, in microcosm experiments. In the publication, I show that this method, combined with a very simple experimental design, is capable of accurately identifying oil-degrading bacterial species.

Microbial consortia and dynamics of microbial communities

In collaboration with Sarah Preheim, I developed Mystic, an ODE model that simulated the chemical and microbial structure and dynamics in Mystic Lake, a polluted eutrophic dimictic lake outside Boston. In the same study, I designed a simple algorithm to identify groups of microbial species (more truly, 16S-amplicon OTUs) that had nearly identical spatial distributions in the lake. This result reproduced known ecological relations and predicts many more. Combined with a single-cell assay, our results suggested that the lake is full of previously-unknown microbial consortia.

Clinical trial design

Fecal microbiota transplant (FMT), the wholesale replacement of a sick person’s gut microbiota with a healthy donor’s, is an extremely effective treatment for often difficult-to-treat Clostridium difficile infections. Gastroenterologists have begun experimenting with using FMT to treat other difficult-to-treat diseases that may involve the gut microbiota, notably inflammatory bowel disease (IBD). Unfortunately, FMT-for-IBD trials have produced mixed results, which may have been due to the fact that only some stool donors produce stool that causes patients to respond to FMT. In theoretical work, I helped derive power calculations that account for this “donor effect”. We also developed adaptive trial strategies that can dramatically improve power.

Skin microbiome

I am interested in the dynamics of the human skin microbiome as it relates to the structure of the skin and perturbations to the microbial environment, like hand-washing, use of hand sanitizer, and antibiotic administration. Understanding these dynamics might inform the rational design of a microbiome transplant, which might help treat diseases of external tissues like acne, swimmer’s ear, sinus infections, and eczema.

I wrote my thesis proposal (download) on this topic. I think these are great ideas but I was not able to pursue them during my PhD. If you find any of those ideas interesting, let me know!


In collaboration with the Hazen Lab at the University of Tennessee (Knoxville) and other members of the DOE-funded ENIGMA project, I designed an experiment to measure the effect of repeated carbon amendments in a natural groundwater ecosystem. The field studies will be performed at Oak Ridge National Laboratory, where the groundwater is contaminated with uranium and other byproducts of WWII nuclear weapons fabrication.

One way to manage uranium-contaminated groundwater is to add carbon. To extract energy from this carbon, bacteria respire using uranium. The reduced uranium is insoluble and thus relatively immobile and at a much smaller risk of flowing into human drinking water sources. There is anecdotal evidence and some previous literature (download my lit review) that suggests that repeatedly adding carbon could be more effective at immobilizing uranium than a single, large dose. This project will explore that possibility.