Integrating experiments and AI to outpace antibiotic resistance:
from core principles to emergent properties
Antibiotic resistance is a defining 21st‑century challenge. We believe that breakthroughs in artificial intelligence, genetic sequencing, and robotic automation will enable meeting this challenge from directions that were unimaginable a decade ago. Our research combines experiments and computation to uncover the principles governing drug–bacteria–host interactions and to translate these principles into new treatment strategies. We discover antibacterial lead compounds with unconventional mechanisms of action and map how changes in bacterial metabolism can be harnessed to augment the efficacy of existing antibiotics. In parallel, we are investigating the rules that stabilize or destabilize bacteria–bacteria interactions in resilient communities that support health, and how determinants of bacteria colonization in different host niches, from the digestive tract to solid cancer tumors.🧪+🧑🏻💻+ 🤖 = 🙀🙀🙀
AI-guided discovery of new antibiotics
The rapid rise of antimicrobial resistance is eroding the effectiveness of frontline antibiotics. Yet most recent efforts in drug development have focused on producing derivatives of known antibiotic classes rather than agents with genuinely new mechanisms of action. We combine robotic automation with in-house machine-learning models to build high-throughput discovery pipelines that mine large compound libraries for unconventional antibacterial activities. This strategy reveals leads with orthogonal toxicity mechanisms—promising starting points for future antibiotics that complement or bypass existing drug classes.🔭 + 🤖 = 💎
Leveraging metabolism to maximize antibiotic efficacy
Bacterial metabolic state strongly impacts sensitivity to antibiotics. Mapping these dependencies is essential for optimizing current therapies and for designing small molecules that steer metabolism to potentiate antibiotic treatment. We develop data-driven approaches that pair high-throughput experimentation with computational inference to discover new drug–metabolism interactions and to uncover their mechanistic underpinnings. These studies highlight actionable cellular processes that can be targeted to boost the efficacy of existing antibiotics.(🧪+🦠)+ 🍎 = 😵
Interactions within microbial communities
Resilient bacterial communities are governed by both cooperation and conflict. We study positive interactions, such as the rules that enable robust cross-feeding in multi-member consortia, as well as negative interactions, including the secretion of toxins used by bacteria to suppress competitors. By integrating controlled consortia with quantitative readouts, we identify design principles that stabilize beneficial exchanges while accounting for antagonism that shapes community structure.🦠+🦠+🦠 = 🥳
Evolutionary adaptation in the tumor microbiome
The human microbiome has emerged as a major player in cancer biology. Recent studies revealed clinically relevant associations between human microbiota and therapy success, and have identified some of the mechanisms facilitating these interactions. Research on patient tumors suggests that many tumors harbor their own microbiome. Yet despite excitement about the tumor microbiome, key questions still remain unanswered. Using a microbial-centric view, we study how bacteria infect and colonize solid tumors and how they evolutionarily adapt to the unique conditions of the tumor microenvironment. We address these questions by using both model bacterial species and clinical isolates cultured directly from tumors resected from human patients. We complement in vitro studies with mouse models of bacteria-colonized tumors.🐭+🧪+🦠 > 🐭+🧪
Science education and outreach
Our lab is highly engaged with science education and outreach directed towards wide and diverse audiences. Within the Graduate School of Biomedical Sciences, Amir is core faculty (developer and lecturer) of two academic courses and has taught multiple workshops on advanced research topics. Additional details can be found on the Teaching page of this site. Importantly, our commitment to science education extends far beyond the community of the medical school.To engage remote audiences in scientific outreach, our lab developed a unique robotic platform that allows remote students to perform empirical experiments without setting foot on campus. We used this platform to engage hundreds of high-school students in a multi-week lab-evolution experiment on the emergence of superbugs (multidrug-resistant strains). Additional details can be found on the Outreach page of this site.
