My research aims to move microbiome science from broad compositional descriptions to a strain-resolved, quantitative understanding of microbial ecology and evolution in human health and disease. I aim to combine long-read metagenomics, synthetic biology, and computational modeling to connect macro-level phenotypes (such as disease progression or therapy response) back to specific microbial strains and alleles.
Microbiome science
1. Long-read metagenomics and strain-level ecology
I develop long-read metagenomic methods that enable the study of microbial strains and alleles over time, greatly improving upon short-read techniques which only resolve to higher levels of taxonomic identification. I apply long-read metagenomics to cross-sectional and longitudinal cohorts to understand how the gut microbiome changes during disease and therapeutic intervention. Furthermore, I demonstrate specific improvements in genome recovery between long-read metagenomics compared with short-read and synthetic long-read approaches.
2. Monitoring of pathogens and antibiotic resistance genes in wastewater
I employ a combination of long-read metagenomics, shotgun metagenomics, and meta-transcriptomics to enable mapping the presence and genomic context of pathogens and antibiotic resistance genes in wastewater. These efforts aim to create a framework where strain- and allele-level measurements can detect rising levels of pathogens against a complex microbial background, and detect potentially multi-drug resistant species or strains.
3. Engineered microbial therapeutics, fermentation, and sensing
Building on my earlier work in synthetic biology, I use sequencing approaches to design and verify the behavior of engineered bacteria and microbial communities in situ. I am interested in how synthetic gene circuits and consortia behave under realistic ecological and evolutionary pressures, and how we can use this knowledge to build robust intratumoral therapies, probiotic consortia, and microbial sensors.
4. Machine learning and genome-informed models
I am also interested in using high-quality, long-read–derived genomes from diverse microbiomes as training data for DNA-level machine learning models, with the goal of improving functional prediction from sequence and informing the design of more stable engineered systems.
Synthetic gene circuits, engineered consortia, and microelectronics
Before focusing on microbiome-wide, strain-resolved methods, much of my work centered on engineering genetic circuits in bacteria and understanding their behavior from single cells to ecological communities:
1. Synchronized lysis circuit (SLC) for therapeutic delivery
I developed the synchronized lysis circuit (SLC), which couples bacterial quorum sensing to phage protein–mediated lysis. At high cell density, SLC bacteria lyse synchronously, releasing intracellular contents (including therapeutic proteins) and attenuating the population. This circuit was tested in mouse models of cancer and demonstrated the feasibility of using engineered bacteria to deliver therapeutics inside tumors.
2. Stability and evolution of genetic circuits
I studied how gene circuits evolve and fail over time, and how ecological design can stabilize their behavior. This work included engineering synthetic microbial ecosystems—such as rock–paper–scissors dynamics and quorum-regulated population control—that maintain circuit function and reduce the emergence of cheaters.
3. Single-cell protein expression and circuit dynamics
Engineered circuits provide a way to explore the landscape of possible protein expression states at the single-cell level. I used microfluidics and time-lapse microscopy to connect single-cell dynamics to population-level behavior and to inform computational models at the single-cell level. This information may help inform the design of future circuits, and possibly the behaviors that manifest in nature.
4. Interfacing genetic circuits with electrical circuits
I helped develop platforms where bacterial population dynamics are read out using electrochemical impedance spectroscopy. Population control circuits were used to make bacteria “communicate” with electrodes by modulating the conductivity of their environment, providing a path toward hybrid computational devices and portable continuous biosensors that couple gene circuit activity to an electrical signal.
