Robert Heeter

Elucidating the structures of unknown small molecules remains a major challenge in chemistry, pharmacology, and drug discovery. While nuclear magnetic resonance (NMR) spectroscopy is the gold standard for structure elucidation, interpreting spectra remains labor-intensive and requires expert knowledge to navigate the vast space of possible structures. Here, we present an AI system for rapid, automated de novo structure elucidation that combines a molecular generative model with a scientific reasoning agent. Its generative component, C hef NMR2, is an atomic flow-matching model that predicts small-molecule structures from molecular formulae and one-dimensional ¹ H and ¹³ C NMR spectra. ChefNMR2 - Agen t augments our generative model with large language models (LLMs) to reason over and refine candidate structures. Across diverse benchmarks, ChefNMR2 substantially outperforms existing tools, offering calibrated confidence estimates and two to three orders of magnitude faster inference than frontier LLMs. We use ChefNMR2 - Agen t to elucidate the structures of four novel natural products, matching human-chemist accuracy in a fraction of the time. Our work demonstrates how specialized generative models and general-purpose reasoning agents can unite to solve challenging scientific inverse problems and accelerate NMR structure elucidation, thus opening new possibilities for AI-driven small-molecule discovery across diverse disciplines.

The ability of cryo-electron microscopy (cryo-EM) to interrogate the atomic structure and motion of biomolecules has motivated the development of a wide range of algorithms for heterogeneity analysis. However, evaluating and comparing these methods remains challenging because ground-truth structures are generally unknown for experimental samples. Here, we introduce the 2026 Community-Wide Assessment of Heterogeneous Reconstruction Algorithms (CAHRA), a community-wide challenge centered on three benchmark datasets that probe distinct tasks in heterogeneity analysis. These include (1) compositional heterogeneity arising from mixtures of distinct protein complexes, (2) continuous conformational variability and atomic modeling, and (3) entanglement between molecular conformation and image pose. We describe the design, construction, and validation of these datasets, as well as their use in the recently completed CAHRA Challenge.

Continuous bladder irrigation (CBI) is a frequent postoperative urological procedure that continuously flushes the bladder with saline. Its main goal is to control gross hematuria (blood in urine) and prevent the formation of clots that can block urinary flow and extend hospital stays. Despite its widespread use, traditional CBI methods are flawed. Currently, there is no standardized or quantitative method to evaluate the severity of hematuria; instead, clinicians rely on subjective visual assessments, using imprecise color descriptors such as "rose" or "cherry red." Saline flow rates must also be adjusted manually, requiring frequent bedside monitoring to assess urine color, regulate inflow, and replace fluid bags. This labor-intensive and inconsistent process contributes to complications in roughly 50% of patients receiving CBI. While some research efforts have explored optical sensing or automated flow regulation, none have successfully delivered a comprehensive solution that combines measurement, automation, alerts, and a user interface in a system suitable for clinical use. To overcome these challenges, we created UroFlo: a smart, adaptive CBI platform designed to streamline and improve hematuria management. UroFlo integrates five essential features: 1) quantitative hematuria analysis; 2) automated inflow control; 3) real-time monitoring of supply and waste volumes; 4) automated caregiver notifications; and 5) an intuitive user interface. By combining objective data with automated decision-making and intel-ligent alerts, UroFlo reduces the need for constant supervision, ensures consistency across care teams, and improves patient outcomes. This system represents a significant advancement in CBI technology, setting a new benchmark for standardizing care and enhancing safety.