The Margaux Pinney lab. Photo: Justin Wang.
Margaux Pinney is a biochemist whose lab utilizes advanced microfluidic technologies to analyze thousands of protein variants simultaneously — measuring catalytic rates, substrate affinities, and inhibitor sensitivities with unprecedented scale and precision.
From understanding how mutations drive drug resistance and disease to discovering the general principles of protein evolution, the Pinney lab seeks to accelerate the design of therapeutic enzymes and inhibitors.
QB3-Berkeley: What's a question or challenge that your lab is trying to solve?
Margaux Pinney: Life depends on enzymes, which accelerate nearly every chemical reaction in the cell. A central goal of enzymology has long been to predict, manipulate, and ultimately design enzyme function from sequence. My lab addresses this challenge by asking how enzyme sequence determines quantitative catalytic behavior across different environments.
Genomic sequencing has revealed billions of sequences, and deep learning methods like AlphaFold enable the prediction of static three-dimensional structures. Nevertheless, we still cannot reliably predict how fast an enzyme catalyzes a reaction, how tightly it binds substrates, or how mutations alter these properties. That gap limits our ability to understand enzyme adaptation, interpret disease mutations, engineer useful enzymes, and anticipate drug resistance.
The challenge is scale. Classical enzymology was built for detailed dissection of up to tens of proteins, not for mapping thousands of sequences under matched conditions. My lab addresses this by leveraging high-throughput microfluidic platforms. With HT-MEK, our High-Throughput Microfluidic Enzyme Kinetics platform, we can express, purify, and quantitatively assay up to approximately 1,800 enzyme variants on a single device, measuring parameters such as catalytic turnover, efficiencies, inhibitor sensitivity, and binding affinity.
QB3: How does your lab use high-throughput microfluidics to reveal how enzymes function?
MP: For adenylate kinase, we have used HT-MEK to build a growing dataset of catalytic constants for hundreds, and soon thousands, of naturally occurring sequences. The data revealed patterns that only become visible at scale: closely related enzymes can behave very differently, and individual mutations can strongly alter catalytic rates.
We show that protein language models are good at organizing enzyme sequences by structural types, but that organization did not translate into accurate predictions of catalytic behavior. To better predict properties such as catalytic rate, models need experimentally measured high-throughput kinetic data.
These approaches also apply to drug discovery. When a pathogen encounters an inhibitor, it explores the same sequence-function landscape, selecting for mutations that preserve activity while evading the drug. If we can map that landscape in advance, we can design inhibitors that are harder to escape.
The same approach can reveal cryptic or allosteric sites: hidden regions outside the active site that strongly affect catalysis and inhibitor sensitivity, offering the first step toward targeting new functional vulnerabilities.
QB3: What defines your lab culture and environment?
MP: My lab sits at the intersection of quantitative enzymology, biophysics, microfluidics, evolutionary biochemistry, bioengineering, and machine learning — and I think of that breadth as a feature. Trainees come from Chemistry, Biophysics, MCB, Bioengineering, and related fields, and we try to build a shared fluency across those areas.
One major training emphasis is experimental design and data analysis. I want every person who comes through my lab to be able to independently generate and interpret large biochemical datasets — to have strong instincts for what a measurement means and where it can go wrong.
Our projects often involve building new tools, from microfluidic methods to computational pipelines, so trainees gain experience both constructing technologies and deploying them on fundamental biological questions. That combination prepares people to work at the frontier.
I also share all datasets and software openly to foster a culture where the goal is to advance science collectively. Every alum from my lab so far has stayed in science — moving on to PhD, MD, and MD/PhD programs, or research roles in industry. That continuity matters to me. I want the people who train with me to feel like scientists when they leave, regardless of where they go next.
QB3: Which mentors inspired your scientific journey?
MP: Two mentors shaped me fundamentally. My graduate advisor, Dan Herschlag at Stanford, taught me to ask mechanistic questions precisely — not to be satisfied with phenomenological descriptions when a physical or chemical explanation is within reach. Dan modeled what it looks like to do rigorous science and never be satisfied with incremental work. That rigor is something I try to bring to everything my lab does.
Polly Fordyce, my postdoctoral mentor at Stanford, gave me the engineering mindset. Working in Polly's lab is where I learned microfluidics and fell in love with the idea of not waiting for a method, assay, or instrument to be developed to answer a question, but of building it. I learned to love tinkering and to treat technology development as a direct path to answering biological questions that could not otherwise be asked.
QB3: How do collaboration and interdisciplinary research influence your work?
MP: Collaboration is central to most of the science my lab does. Our recent paper (Muir et al.) is a good example: it required high-throughput experimental biochemistry from my lab, machine and deep-learning expertise from the Marks and Keiser groups, and a shared commitment to understanding where existing computational models were falling short. None of us could have done that work alone, and the scientific question itself emerged from the collaboration rather than preceding it.
Interdisciplinary work surfaces questions you would not have thought to ask from inside a single field. My Master of Science in Medicine degree at Stanford covered the first two pre-clinical years of the M.D. curriculum, giving me a working fluency in human physiology and disease. It shapes which enzymes we choose to study and allows me to have deeper conversations with physician-scientists about what matters clinically. Building that kind of common language has consistently pulled my research in unexpected directions.