Hyojin Kim
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Hyojin Kim has a total of 50 co-authors. Here are the top co-authors by number of shared publications:
- Derek Jones
University of California San Diego, La Jolla, CA, USA
2 shared publications Garrett Stevenson
Computational Engineering Division, Lawrence Livermore National Laboratory, Livermore, CA, United States
2 shared publicationsJonathan E. Allen
Global Security Computing Applications Division, Lawrence Livermore National Laboratory, Livermore, CA, United States
2 shared publications- Jonathan E. Allen
Global Security Computing Applications Division, Lawrence Livermore National Laboratory, Livermore, CA, USA
2 shared publications Kyle Champley
Nondestructive Characterization Institute, Lawrence Livermore National Laboratory, Livermore, CA, USA
2 shared publicationsStewart He
Global Security Computing Applications Division, Lawrence Livermore National Laboratory, Livermore, CA, United States
2 shared publications
Latest Publications
- Open Access
- 3 August 2025
- Published by American Chemical Society (ACS) in ChemRxiv
ChemBERTa-3: An Open Source Training Framework for Chemical Foundation Models
The rapid advancement of machine learning in computational chemistry has opened new doors for designing molecules, predicting molecular properties, and discovering novel materials. However, building scalable and robust models for molecular property prediction remains a significant challenge due to the vast size and complexity of chemical space. In this paper, we introduce ChemBERTa-3, an open-source training framework designed to train and fine-tune large-scale chemical foundation models. We explore the potential of multiple model architectures by evaluating their performance across various molecular datasets from the MoleculeNet suite. Our experiments demonstrated that pre-training on the expansive ZINC20 dataset yields models capable of performing well on both classification and regression tasks, providing valuable insights into drug discovery and materials science. For scalability, we leveraged both AWS-based Ray deployments and on-premise high-performance computing clusters to support the processing power required to train on billions of molecules. In support of reproducible and extensible science, we have open-sourced all ChemBERTa3 models.
- 3 August 2025
- Published by American Chemical Society (ACS) in ChemRxiv
Open AccessChemBERTa-3: An Open Source Training Framework for Chemical Foundation Models
The rapid advancement of machine learning in computational chemistry has opened new doors for designing molecules, predicting molecular properties, and discovering novel materials. However, building scalable and robust models for molecular property prediction remains a significant challenge due to the vast size and complexity of chemical space. In this paper, we introduce ChemBERTa-3, an open-source training framework designed to train and fine-tune large-scale chemical foundation models. We explore the potential of multiple model architectures by evaluating their performance across various molecular datasets from the MoleculeNet suite. Our experiments demonstrated that pre-training on the expansive ZINC20 dataset yields models capable of performing well on both classification and regression tasks, providing valuable insights into drug discovery and materials science. For scalability, we leveraged both AWS-based Ray deployments and on-premise high-performance computing clusters to support the processing power required to train on billions of molecules. In support of reproducible and extensible science, we have open-sourced all ChemBERTa3 models.
- 3 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 11  (25730436) , 362-376
Distributed Stochastic Optimization of a Neural Representation Network for Time-Space Tomography Reconstruction
4D time-space reconstruction of dynamic events or deforming objects using X-ray computed tomography (CT) is an important inverse problem in non-destructive evaluation. Conventional back-projection based reconstruction methods assume that the object remains static for the duration of several tens or hundreds of X-ray projection measurement images (reconstruction of consecutive limited-angle CT scans). However, this is an unrealistic assumption for many in-situ experiments that causes spurious artifacts and inaccurate morphological reconstructions of the object. To solve this problem, we propose to perform a 4D time-space reconstruction using a distributed implicit neural representation (DINR) network that is trained using a novel distributed stochastic training algorithm. Our DINR network learns to reconstruct the object at its output by iterative optimization of its network parameters such that the measured projection images best match the output of the CT forward measurement model. We use a forward measurement model that is a function of the DINR outputs at a sparsely sampled set of continuous valued 4D object coordinates. Unlike previous neural representation architectures that forward and back propagate through dense voxel grids that sample the object's entire time-space coordinates, we only propagate through the DINR at a small subset of object coordinates in each iteration resulting in an order-of-magnitude reduction in memory and compute for training. DINR leverages distributed computation across several compute nodes and GPUs to produce high-fidelity 4D time-space reconstructions. We use both simulated parallel-beam and experimental cone-beam X-ray CT datasets to demonstrate the superior performance of our approach.
- 3 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Imaging
- Vol. 11  (25730436) , 362-376
Distributed Stochastic Optimization of a Neural Representation Network for Time-Space Tomography Reconstruction
4D time-space reconstruction of dynamic events or deforming objects using X-ray computed tomography (CT) is an important inverse problem in non-destructive evaluation. Conventional back-projection based reconstruction methods assume that the object remains static for the duration of several tens or hundreds of X-ray projection measurement images (reconstruction of consecutive limited-angle CT scans). However, this is an unrealistic assumption for many in-situ experiments that causes spurious artifacts and inaccurate morphological reconstructions of the object. To solve this problem, we propose to perform a 4D time-space reconstruction using a distributed implicit neural representation (DINR) network that is trained using a novel distributed stochastic training algorithm. Our DINR network learns to reconstruct the object at its output by iterative optimization of its network parameters such that the measured projection images best match the output of the CT forward measurement model. We use a forward measurement model that is a function of the DINR outputs at a sparsely sampled set of continuous valued 4D object coordinates. Unlike previous neural representation architectures that forward and back propagate through dense voxel grids that sample the object's entire time-space coordinates, we only propagate through the DINR at a small subset of object coordinates in each iteration resulting in an order-of-magnitude reduction in memory and compute for training. DINR leverages distributed computation across several compute nodes and GPUs to produce high-fidelity 4D time-space reconstructions. We use both simulated parallel-beam and experimental cone-beam X-ray CT datasets to demonstrate the superior performance of our approach.
- Open Access
- 3 January 2025
- Published by Frontiers Media SA in Frontiers in Pharmacology
- Vol. 15, 1518875
Protein-ligand binding affinity prediction using multi-instance learning with docking structures
Introduction: Recent advances in 3D structure-based deep learning approaches demonstrate improved accuracy in predicting protein-ligand binding affinity in drug discovery. These methods complement physics-based computational modeling such as molecular docking for virtual high-throughput screening. Despite recent advances and improved predictive performance, most methods in this category primarily rely on utilizing co-crystal complex structures and experimentally measured binding affinities as both input and output data for model training. Nevertheless, co-crystal complex structures are not readily available and the inaccurate predicted structures from molecular docking can degrade the accuracy of the machine learning methods.Methods: We introduce a novel structure-based inference method utilizing multiple molecular docking poses for each complex entity. Our proposed method employs multi-instance learning with an attention network to predict binding affinity from a collection of docking poses.Results: We validate our method using multiple datasets, including PDBbind and compounds targeting the main protease of SARS-CoV-2. The results demonstrate that our method leveraging docking poses is competitive with other state-of-the-art inference models that depend on co-crystal structures.Discussion: This method offers binding affinity prediction without requiring co-crystal structures, thereby increasing its applicability to protein targets lacking such data.
- 3 January 2025
- Published by Frontiers Media SA in Frontiers in Pharmacology
- Vol. 15, 1518875
Open AccessProtein-ligand binding affinity prediction using multi-instance learning with docking structures
Introduction: Recent advances in 3D structure-based deep learning approaches demonstrate improved accuracy in predicting protein-ligand binding affinity in drug discovery. These methods complement physics-based computational modeling such as molecular docking for virtual high-throughput screening. Despite recent advances and improved predictive performance, most methods in this category primarily rely on utilizing co-crystal complex structures and experimentally measured binding affinities as both input and output data for model training. Nevertheless, co-crystal complex structures are not readily available and the inaccurate predicted structures from molecular docking can degrade the accuracy of the machine learning methods.Methods: We introduce a novel structure-based inference method utilizing multiple molecular docking poses for each complex entity. Our proposed method employs multi-instance learning with an attention network to predict binding affinity from a collection of docking poses.Results: We validate our method using multiple datasets, including PDBbind and compounds targeting the main protease of SARS-CoV-2. The results demonstrate that our method leveraging docking poses is competitive with other state-of-the-art inference models that depend on co-crystal structures.Discussion: This method offers binding affinity prediction without requiring co-crystal structures, thereby increasing its applicability to protein targets lacking such data.
- 4 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Visualization and Computer Graphics
- Vol. 31  (4) , 2196-2210
Bimodal Visualization of Industrial X-Ray and Neutron Computed Tomography Data
Advanced manufacturing creates increasingly complex objects with material compositions that are often difficult to characterize by a single modality. Our collaborating domain scientists are going beyond traditional methods by employing both X-ray and neutron computed tomography to obtain complementary representations expected to better resolve material boundaries. However, the use of two modalities creates its own challenges for visualization, requiring either complex adjustments of bimodal transfer functions or the need for multiple views. Together with experts in nondestructive evaluation, we designed a novel interactive bimodal visualization approach to create a combined view of the co-registered X-ray and neutron acquisitions of industrial objects. Using an automatic topological segmentation of the bivariate histogram of X-ray and neutron values as a starting point, the system provides a simple yet effective interface to easily create, explore, and adjust a bimodal visualization. We propose a widget with simple brushing interactions that enables the user to quickly correct the segmented histogram results. Our semiautomated system enables domain experts to intuitively explore large bimodal datasets without the need for either advanced segmentation algorithms or knowledge of visualization techniques. We demonstrate our approach using synthetic examples, industrial phantom objects created to stress bimodal scanning techniques, and real-world objects, and we discuss expert feedback.
- 4 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Visualization and Computer Graphics
- Vol. 31  (4) , 2196-2210
Bimodal Visualization of Industrial X-Ray and Neutron Computed Tomography Data
Advanced manufacturing creates increasingly complex objects with material compositions that are often difficult to characterize by a single modality. Our collaborating domain scientists are going beyond traditional methods by employing both X-ray and neutron computed tomography to obtain complementary representations expected to better resolve material boundaries. However, the use of two modalities creates its own challenges for visualization, requiring either complex adjustments of bimodal transfer functions or the need for multiple views. Together with experts in nondestructive evaluation, we designed a novel interactive bimodal visualization approach to create a combined view of the co-registered X-ray and neutron acquisitions of industrial objects. Using an automatic topological segmentation of the bivariate histogram of X-ray and neutron values as a starting point, the system provides a simple yet effective interface to easily create, explore, and adjust a bimodal visualization. We propose a widget with simple brushing interactions that enables the user to quickly correct the segmented histogram results. Our semiautomated system enables domain experts to intuitively explore large bimodal datasets without the need for either advanced segmentation algorithms or knowledge of visualization techniques. We demonstrate our approach using synthetic examples, industrial phantom objects created to stress bimodal scanning techniques, and real-world objects, and we discuss expert feedback.
- Open Access
DOLCE: A Model-Based Probabilistic Diffusion Framework for Limited-Angle CT Reconstruction
Limited-Angle Computed Tomography (LACT) is a non-destructive evaluation technique used in a variety of applications ranging from security to medicine. The limited angle coverage in LACT is often a dominant source of severe artifacts in the reconstructed images, making it a challenging inverse problem. We present DOLCE, a new deep model-based framework for LACT that uses a conditional diffusion model as an image prior. Diffusion models are a recent class of deep generative models that are relatively easy to train due to their implementation as image denoisers. DOLCE can form high-quality images from severely under-sampled data by integrating data-consistency updates with the sampling updates of a diffusion model, which is conditioned on the transformed limited-angle data. We show through extensive experimentation on several challenging real LACT datasets that, the same pre-trained DOLCE model achieves the SOTA performance on drastically different types of images. Additionally, we show that, unlike standard LACT reconstruction methods, DOLCE naturally enables the quantification of the reconstruction uncertainty by generating multiple samples consistent with the measured data.
Open AccessDOLCE: A Model-Based Probabilistic Diffusion Framework for Limited-Angle CT Reconstruction
Limited-Angle Computed Tomography (LACT) is a non-destructive evaluation technique used in a variety of applications ranging from security to medicine. The limited angle coverage in LACT is often a dominant source of severe artifacts in the reconstructed images, making it a challenging inverse problem. We present DOLCE, a new deep model-based framework for LACT that uses a conditional diffusion model as an image prior. Diffusion models are a recent class of deep generative models that are relatively easy to train due to their implementation as image denoisers. DOLCE can form high-quality images from severely under-sampled data by integrating data-consistency updates with the sampling updates of a diffusion model, which is conditioned on the transformed limited-angle data. We show through extensive experimentation on several challenging real LACT datasets that, the same pre-trained DOLCE model achieves the SOTA performance on drastically different types of images. Additionally, we show that, unlike standard LACT reconstruction methods, DOLCE naturally enables the quantification of the reconstruction uncertainty by generating multiple samples consistent with the measured data.
- Open Access
- 21 April 2022
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 62  (10) , 2301-2315
Pose Classification Using Three-Dimensional Atomic Structure-Based Neural Networks Applied to Ion Channel–Ligand Docking
The identification of promising lead compounds showing pharmacological activities toward a biological target is essential in early stage drug discovery. With the recent increase in available small-molecule databases, virtual high-throughput screening using physics-based molecular docking has emerged as an essential tool in assisting fast and cost-efficient lead discovery and optimization. However, the best scored docking poses are often suboptimal, resulting in incorrect screening and chemical property calculation. We address the pose classification problem by leveraging data-driven machine learning approaches to identify correct docking poses from AutoDock Vina and Glide screens. To enable effective classification of docking poses, we present two convolutional neural network approaches: a three-dimensional convolutional neural network (3D-CNN) and an attention-based point cloud network (PCN) trained on the PDBbind refined set. We demonstrate the effectiveness of our proposed classifiers on multiple evaluation data sets including the standard PDBbind CASF-2016 benchmark data set and various compound libraries with structurally different protein targets including an ion channel data set extracted from Protein Data Bank (PDB) and an in-house KCa3.1 inhibitor data set. Our experiments show that excluding false positive docking poses using the proposed classifiers improves virtual high-throughput screening to identify novel molecules against each target protein compared to the initial screen based on the docking scores.
- 21 April 2022
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 62  (10) , 2301-2315
Open AccessPose Classification Using Three-Dimensional Atomic Structure-Based Neural Networks Applied to Ion Channel–Ligand Docking
The identification of promising lead compounds showing pharmacological activities toward a biological target is essential in early stage drug discovery. With the recent increase in available small-molecule databases, virtual high-throughput screening using physics-based molecular docking has emerged as an essential tool in assisting fast and cost-efficient lead discovery and optimization. However, the best scored docking poses are often suboptimal, resulting in incorrect screening and chemical property calculation. We address the pose classification problem by leveraging data-driven machine learning approaches to identify correct docking poses from AutoDock Vina and Glide screens. To enable effective classification of docking poses, we present two convolutional neural network approaches: a three-dimensional convolutional neural network (3D-CNN) and an attention-based point cloud network (PCN) trained on the PDBbind refined set. We demonstrate the effectiveness of our proposed classifiers on multiple evaluation data sets including the standard PDBbind CASF-2016 benchmark data set and various compound libraries with structurally different protein targets including an ion channel data set extracted from Protein Data Bank (PDB) and an in-house KCa3.1 inhibitor data set. Our experiments show that excluding false positive docking poses using the proposed classifiers improves virtual high-throughput screening to identify novel molecules against each target protein compared to the initial screen based on the docking scores.
- Top Cited PapersOpen Access
- 23 March 2021
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 61  (4) , 1583-1592
Improved Protein–Ligand Binding Affinity Prediction with Structure-Based Deep Fusion Inference
Predicting accurate protein–ligand binding affinities is an important task in drug discovery but remains a challenge even with computationally expensive biophysics-based energy scoring methods and state-of-the-art deep learning approaches. Despite the recent advances in the application of deep convolutional and graph neural network-based approaches, it remains unclear what the relative advantages of each approach are and how they compare with physics-based methodologies that have found more mainstream success in virtual screening pipelines. We present fusion models that combine features and inference from complementary representations to improve binding affinity prediction. This, to our knowledge, is the first comprehensive study that uses a common series of evaluations to directly compare the performance of three-dimensional (3D)-convolutional neural networks (3D-CNNs), spatial graph neural networks (SG-CNNs), and their fusion. We use temporal and structure-based splits to assess performance on novel protein targets. To test the practical applicability of our models, we examine their performance in cases that assume that the crystal structure is not available. In these cases, binding free energies are predicted using docking pose coordinates as the inputs to each model. In addition, we compare these deep learning approaches to predictions based on docking scores and molecular mechanic/generalized Born surface area (MM/GBSA) calculations. Our results show that the fusion models make more accurate predictions than their constituent neural network models as well as docking scoring and MM/GBSA rescoring, with the benefit of greater computational efficiency than the MM/GBSA method. Finally, we provide the code to reproduce our results and the parameter files of the trained models used in this work. The software is available as open source at https://github.com/llnl/fast. Model parameter files are available at ftp://gdo-bioinformatics.ucllnl.org/fast/pdbbind2016_model_checkpoints/.
- 23 March 2021
- Published by American Chemical Society (ACS) in Journal of Chemical Information and Modeling
- Vol. 61  (4) , 1583-1592
Top Cited PapersOpen AccessImproved Protein–Ligand Binding Affinity Prediction with Structure-Based Deep Fusion Inference
Predicting accurate protein–ligand binding affinities is an important task in drug discovery but remains a challenge even with computationally expensive biophysics-based energy scoring methods and state-of-the-art deep learning approaches. Despite the recent advances in the application of deep convolutional and graph neural network-based approaches, it remains unclear what the relative advantages of each approach are and how they compare with physics-based methodologies that have found more mainstream success in virtual screening pipelines. We present fusion models that combine features and inference from complementary representations to improve binding affinity prediction. This, to our knowledge, is the first comprehensive study that uses a common series of evaluations to directly compare the performance of three-dimensional (3D)-convolutional neural networks (3D-CNNs), spatial graph neural networks (SG-CNNs), and their fusion. We use temporal and structure-based splits to assess performance on novel protein targets. To test the practical applicability of our models, we examine their performance in cases that assume that the crystal structure is not available. In these cases, binding free energies are predicted using docking pose coordinates as the inputs to each model. In addition, we compare these deep learning approaches to predictions based on docking scores and molecular mechanic/generalized Born surface area (MM/GBSA) calculations. Our results show that the fusion models make more accurate predictions than their constituent neural network models as well as docking scoring and MM/GBSA rescoring, with the benefit of greater computational efficiency than the MM/GBSA method. Finally, we provide the code to reproduce our results and the parameter files of the trained models used in this work. The software is available as open source at https://github.com/llnl/fast. Model parameter files are available at ftp://gdo-bioinformatics.ucllnl.org/fast/pdbbind2016_model_checkpoints/.
- Open Access
- 27 August 2020
- Published by Royal Society of Chemistry (RSC) in Nanoscale
- Vol. 12  (37) , 19461-19469
Machine vision-driven automatic recognition of particle size and morphology in SEM images
A comprehensive framework to automatically perform size and morphology recognition of nanoparticles in SEM images in a high-throughput manner.
- 27 August 2020
- Published by Royal Society of Chemistry (RSC) in Nanoscale
- Vol. 12  (37) , 19461-19469
Open AccessMachine vision-driven automatic recognition of particle size and morphology in SEM images
A comprehensive framework to automatically perform size and morphology recognition of nanoparticles in SEM images in a high-throughput manner.
- Open Access
Extreme Few-view CT Reconstruction using Deep Inference
Reconstruction of few-view x-ray Computed Tomography (CT) data is a highly ill-posed problem. It is often used in applications that require low radiation dose in clinical CT, rapid industrial scanning, or fixed-gantry CT. Existing analytic or iterative algorithms generally produce poorly reconstructed images, severely deteriorated by artifacts and noise, especially when the number of x-ray projections is considerably low. This paper presents a deep network-driven approach to address extreme few-view CT by incorporating convolutional neural network-based inference into state-of-the-art iterative reconstruction. The proposed method interprets few-view sinogram data using attention-based deep networks to infer the reconstructed image. The predicted image is then used as prior knowledge in the iterative algorithm for final reconstruction. We demonstrate effectiveness of the proposed approach by performing reconstruction experiments on a chest CT dataset.
Open AccessExtreme Few-view CT Reconstruction using Deep Inference
Reconstruction of few-view x-ray Computed Tomography (CT) data is a highly ill-posed problem. It is often used in applications that require low radiation dose in clinical CT, rapid industrial scanning, or fixed-gantry CT. Existing analytic or iterative algorithms generally produce poorly reconstructed images, severely deteriorated by artifacts and noise, especially when the number of x-ray projections is considerably low. This paper presents a deep network-driven approach to address extreme few-view CT by incorporating convolutional neural network-based inference into state-of-the-art iterative reconstruction. The proposed method interprets few-view sinogram data using attention-based deep networks to infer the reconstructed image. The predicted image is then used as prior knowledge in the iterative algorithm for final reconstruction. We demonstrate effectiveness of the proposed approach by performing reconstruction experiments on a chest CT dataset.