About

We study biology through the lens of chemistry and physics, as the physical and chemical properties of biomolecules enable and constrain what biology can do and how it has evolved. Chemistry and physics are needed to understand and predict biology and to manipulate and engineer biological systems for biomedical and other applications. We focus on questions of how enzymes work, how RNA folds, how proteins recognize RNA, RNA/protein interactions in regulation and control, and the evolution of molecules and molecular interactions.

Research topics

• Computer Science
• Biology
• Physics
• Biochemistry
• Chemistry
• Materials science
• Genetics
• Machine Learning
• Nanotechnology
• Data Mining
• Computational biology
• Biophysics
• Computational chemistry
• Engineering
• Biological system
• Data science
• Biochemical engineering
• Statistical physics
• Nuclear magnetic resonance

Selected publications

• High-throughput biochemical phenotyping of SHP2 variants reveals the molecular basis of diseases and allosteric drug inhibition

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-04-01

  articleOpen access

  Interpreting clinical and functional consequences of genetic variants remains challenging due to limited quantitative biochemical data at scale. We applied high-throughput microfluidic enzyme kinetics to profile 190 clinical variants of SHP2, a protein tyrosine phosphatase linked to developmental disorders and cancers. Through >300,000 reaction progress curves, we derived kinetic and thermodynamic parameters quantifying variant effects on catalysis, autoinhibition, stability, phosphopeptide binding, and drug responses. This multidimensional dataset reveals that dysregulated autoinhibition, rather than altered stability or catalysis, predominantly determines SHP2-associated pathogenesis. Thermodynamic modeling reveals that clinical-stage allosteric inhibitors preferentially stabilize a previously underappreciated, partially active conformation over the fully inactive state, leading to variant-dependent drug responses. Our high-throughput biochemical framework establishes a general approach to decipher the biochemical logic connecting protein variants to clinical outcomes.

  PublisherOA PDFDOI

• Understanding how enzymes work: the journey to ensemble–function studies

  FEBS Journal · 2026-02-18

  articleOpen access1st authorCorresponding

  In this perspective, we describe how we arrived at a framework of ensemble–function analyses to quantitatively dissect enzyme catalysis and biological function more broadly. Serine proteases are described in biochemistry textbooks to illustrate enzyme mechanisms, yet those descriptions do not explain how these enzymes achieve their ~ 10 12 ‐fold rate enhancements. Moving away from the classic descriptions of ‘catalytic triad’ and ‘oxyanion hole’, we returned to the basic physical and chemical interactions in serine protease active sites and identified molecular features that enable a highly efficient reaction path on the enzymes, compared to the uncatalyzed reaction. We then leveraged principles from statistical mechanics to quantify the contributions from each catalytic feature. Combining the contributions from each feature in a ‘catalytic ledger’ provided a quantitative accounting of serine protease catalysis. These analyses revealed previously unrecognized catalytic interactions that are destabilizing in the reaction's ground state—unfavorable bond rotamers, shorter‐than‐ideal distances, and suboptimal hydrogen bonds—each of which is relieved in the transition state, thereby lowering the barrier to reaction. Analogous catalytic features are found in over 30 different protease and nonprotease enzymes spread across 12 structural folds, suggesting that nature has taken advantage of these strategies multiple times in different contexts. In the future, ensemble–function analyses can be used to derive quantitative mechanistic models for other enzymes, to dissect allostery, and to ascertain how molecular machines operate. Ensemble–function also provides a powerful educational approach by linking the complex behavior of biomolecules to the simple chemical and physical principles that are taught in undergraduate classes.

  PublisherOA PDFDOI

• Thermodynamic prediction of RNA cellular activity from sequence via conformational ensembles

  Cell · 2026-03-18

  article
  PublisherDOI

• Code for 'Physics-Grounded Evaluation to Guide Accurate Biomolecular Prediction'

  Code Ocean · 2026-01-01

  otherOpen accessSenior author

  This capsule contains the demo code for the manuscript 'Physics-Grounded Evaluation to Guide Accurate Biomolecular Prediction'. The manuscript is available as a preprint on: https://www.biorxiv.org/content/10.1101/2025.06.30.662466v2 Each demo notebook in /demo is expected to be executed within a minute. The full code associated with this manuscript can be found in Zenodo: 10.5281/zenodo.19139881. The input structures can be found in /structures. The raw output files can be found in /data.

  PublisherDOI

• SEISMICgraph: a web-based tool for RNA structure data visualization

  Nucleic Acids Research · 2025-07-14 · 1 citations

  articleOpen access

  In recent years, RNA has been increasingly recognized for its essential roles in biology, functioning not only as a carrier of genetic information but also as a dynamic regulator of gene expression through its interactions with other RNAs, proteins, and itself. Advances in chemical probing techniques have significantly enhanced our ability to identify RNA secondary structures and understand their regulatory roles. These developments, alongside improvements in experimental design and data processing, have greatly increased the resolution and throughput of structural analyses. Here, we introduce SEISMICgraph, a web-based tool designed to support RNA structure research by offering data visualization and analysis capabilities for a variety of chemical probing modalities. SEISMICgraph enables simultaneous comparison of data across different sequences and experimental conditions through a user-friendly interface that requires no programming expertise. We demonstrate its utility by investigating known and putative riboswitches and exploring how RNA modifications influence their structure and binding. SEISMICgraph's ability to rapidly visualize adenine-dependent structural changes and assess the impact of pseudouridylation on these transitions provides novel insights and establishes a roadmap for numerous future applications.

  PublisherOA PDFDOI

• Quantifying protein unfolding kinetics with a high-throughput microfluidic platform

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-01-18 · 2 citations

  preprintOpen access

  via proteolysis, aggregation, or post-translational modification). Kinetic stability, in addition to thermodynamic stability, can directly impact protein lifetime, abundance, and the formation of alternative, sometimes disruptive states. However, we have very few measurements of protein unfolding rates or how mutations alter these rates, largely due to technical challenges associated with their measurement. To address this need, we developed SPARKfold (Simultaneous Proteolysis Assay Revealing Kinetics of Folding), a microfluidic platform to express, purify, and measure unfolding rate constants for >1000 protein variants in parallel via on-chip native proteolysis. To demonstrate the power and potential of SPARKfold, we determined unfolding rate constants for 1,104 protein samples in parallel. We built a library of 31 dihydrofolate reductase (DHFR) orthologs with up to 78 chamber replicates per variant to provide the statistical power required to evaluate the system's ability to resolve subtle effects. SPARKfold rate constants for 5 constructs agreed with those obtained using traditional techniques across a 150-fold range, validating the accuracy of the technique. Comparisons of mutant kinetic effects via SPARKfold with previously published measurements impacts on folding thermodynamics provided information about the folding transition state and pathways via φ analysis. Overall, SPARKfold enables rapid characterization of protein variants to dissect the nature of the unfolding transition state. In future work, SPARKfold can reveal mutations that drive misfolding and aggregation and enable rational design of kinetically hyperstable variants for industrial use in harsh environments.

  PublisherDOI

• Conformational ensembles reveal the origins of serine protease catalysis

  Science · 2025-02-13 · 38 citations

  articleSenior authorCorresponding

  Enzymes exist in ensembles of states that encode the energetics underlying their catalysis. Conformational ensembles built from 1231 structures of 17 serine proteases revealed atomic-level changes across their reaction states. By comparing the enzymatic and solution reaction, we identified molecular features that provide catalysis and quantified their energetic contributions to catalysis. Serine proteases precisely position their reactants in destabilized conformers, creating a downhill energetic gradient that selectively favors the motions required for reaction while limiting off-pathway conformational states. The same catalytic features have repeatedly evolved in proteases and additional enzymes across multiple distinct structural folds. Our ensemble-function analyses revealed previously unknown catalytic features, provided quantitative models based on simple physical and chemical principles, and identified motifs recurrent in nature that may inspire enzyme design.

  PublisherDOI

• Quantifying Protein Unfolding Kinetics with a High-Throughput Microfluidic Platform

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access
  PublisherDOI

• Physics-Grounded Evaluation to Guide Accurate Biomolecular Prediction

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-07-04 · 6 citations

  preprintOpen accessSenior authorCorresponding

  Abstract Deep learning has revolutionized protein structure prediction, with function prediction on the horizon 1,2 . Biomolecular properties, including structure and all aspects of function, emerge from atomic-level interactions and the probabilities of their formation 3,4 . Learning the physical rules that govern these probabilities allows models to deliver accurate structure predictions and may enable extrapolation beyond the training data—a capability needed to predict the many biologically important functional properties where comprehensive data are not readily attainable 5 . Current structure-based models follow a training logic primarily focused on matching atomic coordinates, rather than atomic interactions and their probabilities. It remains unknown whether the models have learned the physical rules that underlie atomic interactions, the extent of their knowledge, and the prediction errors that arise from limits to this knowledge. We found that state-of-the-art structure prediction models, AlphaFold2, AlphaFold3, and ESMFold, capture basic energetic principles but show pervasive biases in the conformational preferences of molecular interactions. These biases manifest as widespread prediction errors, including the misassignment of a large fraction of side-chain non-covalent interactions—∼30% for the AlphaFold models and ∼60% for ESMFold—and an inability to reproduce experimentally derived conformational ensembles. More than half the errors made by AlphaFold2 and AlphaFold3 are in common, suggesting limitations not overcome by using different model architectures. Overall, our multifaceted, physics-grounded evaluation identified previously unknown, system-wide deficiencies in current structure prediction models. This framework is applicable to and needed for all biomolecular structure and function prediction models that deliver atomic-level structural information. The insights derived from these evaluations will allow researchers to judiciously apply current models and will guide the development of next-generation models to achieve accurate prediction of biomolecular function.

  PublisherOA PDFDOI

• Proteolytic activation of diverse antiviral defense modules in prokaryotes

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-11-14

  preprintOpen access

  Linked protease–effector modules are widespread in prokaryotic antiviral defense, yet the mechanisms of most remain poorly understood. Here we show that four of the most prevalent modules—metallo-β-lactamase (MBL)-fold hydrolase, α/β-hydrolase, Pepco, and EACC1—form latent death effectors that are unleashed by site-specific proteolysis. Genetic, biochemical, and structural analyses reveal novel modes of effector licensing. MBL and α/β-hydrolase are zymogens activated by cleavage at two distinct sites, and upon proteolysis, MBL becomes a Zn 2+ -dependent double-stranded DNA nuclease. In contrast, Pepco and EACC1 act as pore-forming toxins via distinct mechanisms: Pepco constitutively oligomerizes into a denaturation-resistant beta barrel that is activated by cleavage after a specific isoleucine in its C-terminal tail, whereas EACC1 monomers assemble into large membrane pores following removal of an autoinhibitory domain. All four modules are fused to diverse sensors to detect a wide range of phage signals, and EACC1–DnaK chaperone fusions suggest a convergence between defense and general stress responses. These findings establish proteolysisgated activation as a dominant, modular logic for anti-phage defense and reveal parallels with eukaryotic innate immunity.

  PublisherDOI

Recent grants

• Quantitative, High-throughput Mechanistic Enzymology

  NIH · $5.9M · 2002–2027

• NIH Grant R37GM049243

  NIH · $4.2M · 2013

• Fundamental Studies of RNA Conformational Thermodynamics

  NIH · $2.7M · 2019–2024

• Collaborative Research: Systematic Investigation of the Structure, Dynamics, and Energetics of Hydrogen Bonds and the Protein Interior Using Ketosteroid Isomerase and Model Systems

  NSF · $903k · 2017–2023

• NIH Grant P01GM066275

  NIH · $25.1M · 2020

Frequent coauthors

• Thomas R. Cech

  University of Colorado Boulder

  41 shared

• Sebastian Doniach

  Stanford University

  41 shared

• Joseph A. Piccirilli

  University of Chicago

  40 shared

• Patrick O. Brown

  Stanford University

  37 shared

• Suzanne J. Admiraal

  Michigan Medicine

  36 shared

• Rick Russell

  The University of Texas at Austin

  35 shared

• Joël Janin

  Centre National de la Recherche Scientifique

  32 shared

• Dominique Deville‐Bonne

  Sorbonne Université

  32 shared

Labs

• Herschlag LabPI

Education

• B.S., Chemistry

  University of California, Berkeley

  1984

• Ph.D., Chemistry

  Stanford University

  1989

Awards & honors

• Inducted into the American Academy of Arts and Sciences (202…