Search NASA⌕ Search

SEARCH · Search NASA

Results for “next generation sequencing”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 163 records · Page 9
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Quasiparticle Poisoning of Superconducting Qubits from Resonant Absorption of Pair-Breaking Photons

The ideal superconductor provides a pristine environment for the delicate states of a quantum computer: because there is an energy gap to excitations, there are no spurious modes with which the qubits can interact, causing irreversible decay of the quantum state. As a practical matter, however, there exists a high density of excitations out of the superconducting ground state even at ultralow temperature; these are known as quasiparticles. Observed quasiparticle densities are of order 1 μ m - 3 , tens of orders of magnitude greater than the equilibrium density expected from theory. Nonequilibrium quasiparticles extract energy from the qubit mode and can induce dephasing. Here we show that a dominant mechanism for quasiparticle poisoning is direct absorption of high-energy photons at the qubit junction. We use a Josephson junction-based photon source to controllably dose qubit circuits with millimeter-wave radiation, and we use an interferometric quantum gate sequence to reconstruct the charge parity of the qubit. We find that the structure of the qubit itself acts as a resonant antenna for millimeter-wave radiation, providing an efficient path for photons to generate quasiparticles. A deep understanding of this physics will pave the way to realization of next-generation superconducting qubits that are robust against quasiparticle poisoning.

Liu, Chuan-Hong↗
DOE OSTI2023

Emergent layer stacking arrangements in c-axis confined MoTe 2

The layer stacking order in 2D materials strongly affects functional properties and holds promise for next-generation electronic devices. In bulk, octahedral MoTe 2 possesses two stacking arrangements, the ferroelectric Weyl semimetal T d phase and the higher-order topological insulator 1T' phase. However, in thin flakes of MoTe 2 , it is unclear if the layer stacking follows the T d , 1T', or an alternative stacking sequence. Here, we use atomic-resolution scanning transmission electron microscopy to directly visualize the MoTe 2 layer stacking. In thin flakes, we observe highly disordered stacking, with nanoscale 1T' and T d domains, as well as alternative stacking arrangements not found in the bulk. We attribute these findings to intrinsic confinement effects on the MoTe 2 stacking-dependent free energy. Our results are important for the understanding of exotic physics displayed in MoTe 2 flakes. More broadly, this work suggests c-axis confinement as a method to influence layer stacking in other 2D materials.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2021

Accelerating large scale de novo metagenome assembly using GPUs

Metagenomic workflows involve studying uncultured microorganisms directly from the environment. These environmental samples when processed by modern sequencing machines yield large and complex datasets that exceed the capabilities of metagenomic software. The increasing sizes and complexities of datasets make a strong case for exascale-capable metagenome assemblers. However, the underlying algorithmic motifs are not well suited for GPUs. This poses a challenge since the majority of next-generation supercomputers will rely primarily on GPUs for computation. In this paper we present the first of its kind GPU-Accelerated implementation of the local assembly approach that is an integral part of a widely used large-scale metagenome assembler, MetaHipMer. Local assembly uses algorithms that induce random memory accesses and non-deterministic workloads, which make GPU offloading a challenging task. Our GPU implementation outperforms the CPU version by about 7x and boosts the performance of MetaHipMer by 42% when running on 64 Summit nodes.

Awan, Muaaz Gul↗
DOE OSTI2025

In situ Detection of Plasma Induced Surface Interaction based on Deep Learning based Visual Diagnostics (Technical Report)

It is characteristic for many plasma devices to undergo plasma-material interaction leading to surface erosion. These processes, often not easily detectable, lead to changes in device performance and lifespan. State-of-the-art lifetime tests and wear experiments require over 1000s hours. A self-consistent model for accurately predicting the erosion's effects is not available. In situ detection of these processes is not a trivial task since the surface variations at the early stages have a micron scale. Such limitations not only restrict testing and prediction capabilities but also slow the development of new thrusters and limit mission duration. To address these challenges, an in-situ diagnostic for real-time erosion assessment has been developed, aiming to expedite lifetime testing and broaden experimental campaigns. Several works were dedicated to real-time and in situ monitoring of material erosion during plasma exposure using laser holography, microscopy, and with telemicroscopes. However, the applicability of these approaches is limited due to complexity, cost and less flexibility as they often require placing diagnostic equipment inside the vacuum chamber. In collaboration with Princeton Collaborative Research Facility (PCRF), Princeton Plasma Physics Laboratory (PPPL), a new diagnostic approach is developed, where geometry modifications to the ceramic channel walls were introduced that would result in accelerated channel erosion. We employed Long-distance microscope (LDM) imagery, combined with Deep-Learning based Shape from focus or depth from focus (DFF or SFF) approach, that provides an accessible and cost-effective solution. LDM employs focus variation techniques to continuously capture multiple images of the target object at distinct focal planes. DFF, an optical focus variation method, generates a 3D topographical surface depth map from a sequence of variably focused images. Combined with the developed diagnostic, this approach offers a controllable means to study erosion under accelerated conditions. In this work, we develop Neural Network-based DFF algorithm applicable for LDM data to quantitatively evaluate plasma induced surface modification from LDM data. Next, we develop Deep Learning-based super-resolution depth map image reconstruction technique to increase the resolution of depth maps obtained from DFF algorithm to improve the accuracy of erosion measurements. Thirdly, we develop several image processing techniques to remove noise and improve the quality of depth map image. Here we report the results of initial tests for this approach. An experimental setup designed and built in PPPL was employed that consists of a 3-cm gridded ion source that produces a neutralized argon beam with energies up to 600 eV. A hexagonal boron nitride (h-BN) ceramic target, designed based on computational predictions, was used. Tests were conducted to reconstruct the complex geometry of the target under the lighting conditions of the operated ion source.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2025

Turbo‐charging crop improvement: harnessing multiplex editing for polygenic trait engineering and beyond

Multiplex CRISPR editing has emerged as a transformative platform for plant genome engineering, enabling the simultaneous targeting of multiple genes, regulatory elements, or chromosomal regions. This approach is effective for dissecting gene family functions, addressing genetic redundancy, engineering polygenic traits, and accelerating trait stacking and de novo domestication. Its applications now extend beyond standard gene knockouts to include epigenetic and transcriptional regulation, chromosomal engineering, and transgene‐free editing. These capabilities are advancing crop improvement not only in annual species but also in more complex systems such as polyploids, undomesticated wild relatives, and species with long generation times. At the same time, multiplex editing presents technical challenges, including complex construct design and the need for robust, scalable mutation detection. We discuss current toolkits and recent innovations in vector architecture, such as promoter and scaffold engineering, that streamline workflows and enhance editing efficiency. High‐throughput sequencing technologies, including long‐read platforms, are improving the resolution of complex editing outcomes such as structural rearrangements—often missed by standard genotyping—when targeting repetitive or tandemly spaced loci. To fully realize the potential of multiplex genome engineering, there is growing demand for user‐friendly, synthetic biology‐compatible, and scalable computational workflows for gRNA design, construct assembly, and mutation analysis. Experimentally validated inducible or tissue‐specific promoters are also highly desirable for achieving spatiotemporal control. As these tools continue to evolve, multiplex CRISPR editing is poised to become a foundational technology of next‐generation crop improvement to address challenges in agriculture, sustainability, and climate resilience.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2024

Towards reverse mode automatic differentiation of Kokkos-based codes

Derivative computation is a key component of optimization, sensitivity analysis, uncertainty quantification, and the solving of nonlinear problems. Automatic differentiation (AD) is a powerful technique for evaluating such derivatives, and in recent years, has been integrated into programming environments such as Jax, PyTorch, and TensorFlow to support derivative computations needed for training of machine learning models, facilitating wide-spread use of these technologies. The C++ language has become the de facto standard for scientific computing due to numerous factors, yet language complexity has made the wide-spread adoption of AD technologies for C++ difficult, hampering the incorporation of powerful differentiable programming approaches into C++ scientific simulations. This is exacerbated by the increasing emergence of architectures, such as GPUs, with limited memory capabilities and requiring massive thread-level concurrency. C++ AD tools must effectively use these environments to bring novel scientific simulations to next-generation DOE experimental and observational facilities. In this project, we investigated source transformation-based automatic differentiation using LLVM compiler infrastructure to automatically generate portable and efficient gradient computations of Kokkos-based code. We have demonstrated that our proposed strategy is feasible by investigating the usage of a prototype LLVM-based source transformation tool to generate gradients of simple functions made of sequences of simple Kokkos parallel regions. Speedups of up to 500x compared to Sacado were observed on NVIDIA V100 GPU.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

RCSB protein data Bank: Next‐generation advanced search for exploration of experimental structures and computed structure models

Abstract The Protein Data Bank (PDB), established in 1971, is the primary global, open‐access archive for experimentally determined 3D macromolecular structures (proteins, RNA, DNA). The research‐focused RCSB.org web‐portal provides access to these data alongside more than one million machine‐learning‐predicted structure models, greatly expanding the available structural landscape. Rapid growth of both experimental and computational structures has increased the need for powerful yet accessible search tools that serve a broad and diverse scientific community. Herein, we describe a redesigned RCSB Protein Data Bank RCSB.org Advanced Search capability that supports intuitive discovery of 3D structures through a unified interface. This interface integrates annotation‐, sequence‐, and 3D structure‐based searches, embeds an interactive 3D viewer, and incorporates curated biological knowledge, such as catalytic site definitions from Mechanism and Catalytic Site Atlas and ligand‐guided structural motifs, for constructing geometry‐driven queries. A new Chemical Search tool allows definition of chemical queries via an integrated drawing tool or standard identifiers, seamlessly combining them with annotation filters. By allowing query definition directly within spatial and chemical contexts, these search interfaces reduce the need for detailed knowledge of residue numbering, chain identifiers, or external cheminformatics software. This capability enables efficient exploration of structures, chemical diversity, and structure–function relationships across all life domains. The redesigned interfaces can be accessed directly at rcsb.org/search/advanced for Advanced Search and rcsb.org/search/chemical for Chemical Search.

Rose, Yana [Research Collaboratory for Structural ↗
DOE OSTI2023

Multidimensional Simulations of Core Convection

The cores of main sequence intermediate- and high-mass stars are convective. Mixing at the radiative–convective boundary, waves excited by the convection, and magnetic fields generated by convective dynamos all influence the main sequence and post-main sequence evolution of these stars. These effects must be understood to accurately model the structure and evolution of intermediate- and high-mass stars. Unfortunately, there are many challenges in simulating core convection due to the wide range of temporal and spatial scales, as well as many important physics effects. In this review, we describe the latest numerical strategies to address these challenges. We then describe the latest state-of-the-art simulations of core convection, summarizing their main findings. These simulations have led to important insights into many of the processes associated with core convection. Two outstanding problems with multidimensional simulations are, 1. it is not always straightforward to extrapolate from simulation parameters to the parameters of real stars; and 2. simulations using different methods sometimes appear to arrive at contradictory results. To address these issues, next generation simulations of core convection must address how their results depend on stellar luminosity, dimensionality, and turbulence intensity. Furthermore, code comparison projects will be essential to establish robust parameterizations that will become the new standard in stellar modeling.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2026

Automation of Laser Plasma Focused Ion Beam Microscopy for Next-Gen Energy Materials

Automation can revolutionize the use of ultrafast laser ablation and plasma-focused ion beam (PFIB) techniques for high-throughput, reproducible cross-sectioning and various sample preparation in materials characterization. As these methods become essential for analyzing complex energy materials and next-generation devices, efficient, standardized workflows are needed to minimize variability and enhance precision. This work highlights our advancements in developing automated processes for sample preparation that integrates machine learning, workflow optimization, and large-scale data acquisition to improve efficiency and scalability in applications such as electrolyzers, photovoltaic cells, and microelectronics. To streamline cross-sectioning and lamella fabrication, we have implemented fully automated workflows that standardize laser ablation and PFIB milling sequences. These workflows incorporate pre-programmed protocols for material removal, alignment, and thinning, reducing user intervention and ensuring consistency across different sample types. Machine learning algorithms further enhance automation by predicting optimal milling strategies and adapting parameters based on material properties and sectioning requirements. This approach significantly improves throughput while maintaining the structural integrity of prepared samples for high-resolution imaging and analysis, including transmission electron microscopy. Beyond sample preparation, our automation platform enables the acquisition of large, high-resolution datasets through serial sectioning, image alignment, and 3D reconstruction. These automated routines facilitate multi-scale characterization, capturing structural and compositional details from the nanoscale to the device level. By reducing variability and increasing efficiency, our automated approach enhances defect analysis, failure diagnostics, and process optimization, accelerating advancements in materials research and device engineering.

36 MATERIALS SCIENCE↗
DOE OSTI2025

Laser wakefield acceleration driven by a discrete flying focus

Laser wakefield acceleration (LWFA) may enable the next generation of TeV-scale lepton colliders. Reaching such energies will likely require multiple LWFA stages to overcome limitations on the energy gain achievable in a single stage. The use of stages, however, introduces challenges such as alignment, adiabatic matching between stages, and a lower average accelerating gradient. Here, we propose a discrete flying focus that can deliver higher energy gain in a single stage, thereby reducing the number of stages required for a target energy. A sequence of laser pulses with staggered focal points and delays drives a plasma wave in which an electron beam experiences a near-constant accelerating gradient over distances beyond those attainable with a conventional pulse. Simulations demonstrate that a discrete flying focus with a total energy of 150 J can transfer 40 GeV per electron to a 50-pC beam in a single 30-cm stage, corresponding to 50 dephasing lengths.

laser wakefield acceleration↗
DOE OSTI2022

High-throughput detection of T-DNA insertion sites for multiple transgenes in complex genomes

Abstract Background Genetic engineering of crop plants has been successful in transferring traits into elite lines beyond what can be achieved with breeding techniques. Introduction of transgenes originating from other species has conferred resistance to biotic and abiotic stresses, increased efficiency, and modified developmental programs. The next challenge is now to combine multiple transgenes into elite varieties via gene stacking to combine traits. Generating stable homozygous lines with multiple transgenes requires selection of segregating generations which is time consuming and labor intensive, especially if the crop is polyploid. Insertion site effects and transgene copy number are important metrics for commercialization and trait efficiency. Results We have developed a simple method to identify the sites of transgene insertions using T-DNA-specific primers and high-throughput sequencing that enables identification of multiple insertion sites in the T 1 generation of any crop transformed via Agrobacterium . We present an example using the allohexaploid oil-seed plant Camelina sativa to determine insertion site location of two transgenes. Conclusion This new methodology enables the early selection of desirable transgene location and copy number to generate homozygous lines within two generations.

59 BASIC BIOLOGICAL SCIENCES↗
DOE OSTI2021

Surrogate Model of Flexible Research Platform EnergyPlus Models to Enable Sensitivity Analysis

This letter report describes the surrogate models developed from the EnergyPlus model of Oak Ridge National Laboratory’s Flexible Research Platform. Two data-driven black-box models were developed, and the outputs of the surrogate models were compared with the EnergyPlus model. The two models developed are a multilayer perceptron deep learning model, and a long short-term memory (LSTM) neural network model. The three factors for selecting the black-box models are scalability, computation time, and accuracy. A total of 107 input variables were the dominant variables in determining the outputs of building energy consumptions and thermal comfort. A total of 54 output variables were identified as the prediction targets, including the system- and zone-level outputs. The large set of the simulation cases were generated by integrating sensor errors into an emulator based on EnergyPlus and Python EMS, which includes advanced control sequences from ASHRAE Guideline 36-2018: High-Performance Sequences of Operation. The surrogate models were developed based on a set of large-scale simulation runs (i.e., 4,000 runs) on a cloud platform. The comparison analysis shows that the two black-box models had good accuracy for predicting new outputs for sensitivity analysis using the root mean square error metric. As a next step, the developed surrogate models will be used to perform sensitivity analysis for different sensor impacts (e.g., sensor types, sensor locations).

42 ENGINEERING↗
DOE OSTI2024

Tailoring Growth Interfaces of Virtual Substrates for Power Electronics

Power electronics materials are poised to play a critical role in fulfilling next generation energy needs, with up to 90% of future energy demand predicted to flow through power electronics at some point. AlxGa1-xN ranks high among candidate materials, having bipolar dopability, thermal and chemical stability and an ultra-wide bandgap. However, AlGaN growth is limited by a lack of lattice-matched substrates, ultimately stunting material quality at higher thicknesses needed for power electronics applications. Further, high power applications increasingly call for fully vertical device structures, necessitating a conductive substrate. Recently our group identified the (111) plane of TaC as a conductive surface lattice-matched to Al0.55Ga0.45N, taking inspiration from prior work of AlN and GaN binaries on carbide and boride substrates. In this talk we demonstrate the growth of (111)-oriented TaC by RF sputtering. We investigate the interface of TaC with sapphire and SiC substrates and identify means to suppress competing Ta2C nucleation in order to stabilize (111)-oriented TaC. Potential stacking sequences are identified with respect to crystal structure and observed twinning in the TaC films. We next assess structural changes and film recrystallization that results from face-to-face annealing of TaC thin films at high temperatures above 1500 degrees C. Changes to grain structure and domain size are assessed by x-ray diffraction and surface morphology is explored using atomic force microscopy. Figure 1 shows significant improvements to in- and out-of-plane strain following annealing along with the formation of terraced step edges at the film surface. Strain as a function of material composition and thickness is considered, as this may play a major role in future nucleation of AlGaN layers. (1) R. J. in a face-to-face configuration, as illustrated in the schematic at left. Kaplar et al 2017, ECS J. Solid State Sci. Technol. 6 Q3061; (2) D. M. Roberts et al 2022, https://arxiv.org/abs/2208.11769; (3) T. Aizawa et al 2008, J Crys Growth 310, 1 22; (4) R. Liu et al 2002, Appl. Phys. Lett. 81, 3182-3184.

ENGINEERING↗
DOE OSTI2021

An Introduction to Word Embeddings and Language Models

Language models have advanced at a phenomenal pace over the past decade. This document provides a short introduction to terminology, word embeddings (aka low-dimensional representations), and popular large-scale language models (LMs). Word embeddings are used to represent words as numerical vectors and are context-independent, meaning a word can only have a single representation (e.g., club can only be club sandwich, not golf club ). Language models can determine the probability of a given sequence of words occurring in a sentence and can provide context to distinguish between words and phrases that sound similar. LMs are context-dependent (e.g., club can be club sandwich or golf club ) and largely fall in two main classes – autoregressive and autoencoding models. Autoregressive models are pretrained on the classic language modeling task: guess the next token having read all the previous ones. Those models can be fine-tuned and achieve great results on many tasks, the most natural application is text generation. A typical example of such models is GPT, but others include GPT-2, GPT-3, CTLR, TRANSFORMER-XL, REFORMER, XLNET. Autoencoding models are pretrained by corrupting the input tokens in some way and trying to reconstruct the original sentence. They can be fine-tuned and achieve great results on many tasks such as text generation, but their most natural application is sentence classification or token classification. A typical example of such models is BERT, but others include ROBERTA, ALBERT, XML, XML-ROBERTA, FLAUBERT AND LONGFORMER.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Tailoring Growth Interfaces of Virtual Substrates for Power Electronics

Power electronics materials are poised to play a critical role in fulfilling next generation energy needs, with up to 90% of future energy demand predicted to flow through power electronics at some point.[1] Among a number of candidate materials, AlxGa1-xN is the strongest, having bipolar dopability, thermal and chemical stability, an ultra-wide bandgap, and demonstrated experimental feasibility. However, AlGaN growth is limited by a lack of lattice-matched substrates, ultimately stunting material quality at higher thicknesses needed for power electronics applications. Further, high power applications increasingly call for fully vertical device structures, necessitating a conductive substrate. [1] Recently our group identified the (111) plane of TaC as a conductive surface lattice-matched to Al0.55Ga0.45N, taking inspiration from prior work of AlN and GaN binaries on carbide and boride substrates. [2,3,4] In this talk we demonstrate the growth of (111)-oriented TaC by RF sputtering. We investigate the interface of TaC with sapphire and SiC substrates and identify means to suppress competing Ta2C nucleation in order to stabilize (111)-oriented TaC. Potential stacking sequences are identified with respect to crystal structure and observed twinning in the TaC films. We next assess structural changes and film recrystallization that results from face-to-face annealing of TaC thin films at high temperatures above 1500 degrees Celsius. Changes to grain structure and domain size are assessed by x-ray diffraction and surface morphology is explored using atomic force microscopy. Figure 1 shows significant improvements to in- and out-of-plane strain following annealing along with the formation of terraced step edges at the film surface. Strain as a function of material composition and thickness is considered, as this may play a major role in future nucleation of AlGaN layers.

CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND SU↗
DOE OSTI2022

Joint Survey Processing. II. Stellar Proper Motions in the COSMOS Field from Hubble Space Telescope ACS and Subaru Telescope HSC Observations

We analyze stellar proper motions in the COSMOS field to assess the presence of bulk motions. At bright magnitudes (G-band 18.5–20.76 AB), we use the proper motions of 1010 stars in the Gaia DR2 catalog. At the faint end, we compute proper motions of 11,519 pointlike objects at i-band magnitudes 19–25 AB using Hubble ACS and Subaru HSC data, which span two epochs about 11 yr apart. In order to measure these proper motions with unprecedented accuracy at faint magnitudes, we developed a foundational set of astrometric tools that will be required for joint survey processing of data from the next generation of optical/infrared surveys. The astrometric grids of Hubble ACS and Subaru HSC mosaics were corrected at the catalog level using proper motion–propagated and parallax-corrected Gaia DR2 sources. These astrometric corrections were verified using compact extragalactic sources. Upon comparison of our measured proper motions with Gaia DR2, we estimate the uncertainties in our measurements to be ~2–3 mas yr –1 axis –1 , down to 25.5 AB mag. We correct proper motions for the mean motion of the Sun, and we find that late-type main-sequence stars predominantly in the thin disk in the COSMOS field have space velocities mainly toward the Galactic center. We detect candidate high-velocity (≥220 km s –1 ) stars, six of them at ~0.4–6 kpc, from the Gaia sample, and five of them at ~20 kpc, from the faint star HSC and ACS sample. The sources from the faint star sample may be candidate halo members of the Sangarius stream.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

Design, Fabrication, and Screening of Environmental‐Thermal Barrier Coatings Prepared by Ultrafast High‐Temperature Sintering

Abstract The demand for more efficient gas turbines relies heavily on the development of new environmental‐thermal barrier coatings (ETBCs). However, there is still uncertainty about which alloys and composites will be used for the next generation of turbine blades, as well as the most promising coating materials. Herein, an ETBCs development strategy is presented by integrating the coating design, synthesis, and screening using an ultrafast high temperature sintering (UHS) technique to accelerate coating improvements. The initial basis for composition selection is their thermal expansion mismatch with the substrate alloys; for which a temperature‐dependent coefficient of thermal expansion database is created. By combining tape casting method with the UHS technique a high‐throughput synthesis of single and multi‐layer coatings are realized with different compositions, layer stacking sequences, and layer thicknesses. To evaluate the coatings, thermal cycling tests from room temperature to 1300 °C are conducted. The approach enabled coatings on objects with complex geometries, multi‐layer ETBCs, and porosity tailoring by using staged UHS that runs with different temperatures and durations. The fast iteration strategy is more cost‐effective for the screening of ETBCs compared to conventional methods and greater throughput which can be further extended for rapid optimization of other materials systems.

Xie, Hua↗
DOE OSTI2026

A curated benchmark for cofolding models on kinase conformational states

Abstract Protein kinases are critical drug targets, requiring therapeutics that can modulate their active and inactive conformational states. While cofolding models can generate global folds directly from kinase sequences and ligand SMILES strings, these models have not yet been tested on their ability to recover ligand-induced-fit conformational states of the kinase proteins. Here, we introduce KinConfBench, a curated benchmark of 2225 high-quality human kinase chains to evaluate the ability of four state-of-the-art cofolding models—Boltz-2, Chai-1, Protenix, and RoseTTAFold-All-Atom—to recover both canonical and rare conformational states. We show that geometric success metrics of a ligand pose in the active site do not correlate strongly with the correct kinase conformational state, motivating a new set of dynamical benchmarks for assessing cofolding models. While all four cofolding models achieve ~60–80% prediction accuracy for kinase conformational classification, they exhibit severe mode collapse when performing multiple inferences, show negligible structural diversity in sampling induced-fit motions, and display a prevalent “apo-drift” in which most cofolding models predominantly predict the kinase to be in its ligand-free state. Our results highlight that capturing ligand-induced protein conformational diversity, not just geometric fit, is critical for next-generation structure-based drug discovery.

Sun, Kunyang↗