Engineering topics

Gounley, John

Publications and source records attributed to Gounley, John.

At least 19 records

DOE OSTI2024

Enhancing molecular design efficiency: Uniting language models and generative networks with genetic algorithms

This study examines the effectiveness of generative models in drug discovery, material science, and polymer science, aiming to overcome constraints associated with traditional inverse design methods relying on heuristic rules. Generative models generate synthetic data resembling real data, enabling deep learning model training without extensive labeled datasets. They prove valuable in creating virtual libraries of molecules for material science and facilitating drug discovery by generating molecules with specific properties. While generative adversarial networks (GANs) are explored for these purposes, mode collapse restricts their efficacy, limiting novel structure variability. To address this, we introduce a masked language model (LM) inspired by natural language processing. Although LMs alone can have inherent limitations, we propose a hybrid architecture combining LMs and GANs to efficiently generate new molecules, demonstrating superior performance over standalone masked LMs, particularly for smaller population sizes. This hybrid LM-GAN architecture enhances efficiency in optimizing properties and generating novel samples.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Path-BigBird: An AI-Driven Transformer Approach to Classification of Cancer Pathology Reports

PURPOSE Surgical pathology reports are critical for cancer diagnosis and management. To accurately extract information about tumor characteristics from pathology reports in near real time, we explore the impact of using domain-specific transformer models that understand cancer pathology reports. METHODS We built a pathology transformer model, Path-BigBird, by using 2.7 million pathology reports from six SEER cancer registries. We then compare different variations of Path-BigBird with two less computationally intensive methods: Hierarchical Self-Attention Network (HiSAN) classification model and an offthe-shelf clinical transformer model (Clinical BigBird). We use five pathology information extraction tasks for evaluation: site, subsite, laterality, histology, and behavior. Model performance is evaluated by using macro and micro F 1 scores. RESULTS We found that Path-BigBird and Clinical BigBird outperformed the HiSAN in all tasks. Clinical BigBird performed better on the site and laterality tasks. Versions of the Path-BigBird model performed best on the two most difficult tasks: subsite (micro F 1 score of 72.53, macro F 1 score of 35.76) and histology (micro F 1 score of 80.96, macro F 1 score of 37.94). The largest performance gains over the HiSAN model were for histology, for which a Path-BigBird model increased the micro F 1 score by 1.44 points and the macro F 1 score by 3.55 points. Overall, the results suggest that a Path-BigBird model with a vocabulary derived from wellcurated and deidentified data is the best-performing model. CONCLUSION The Path-BigBird pathology transformer model improves automated information extraction from pathology reports. Although Path-BigBird outperforms Clinical BigBird and HiSAN, these less computationally expensive models still have utility when resources are constrained.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2023

Transferring a Molecular Foundation Model for Polymer Property Predictions

Transformer-based large language models have remarkable potential to accelerate design optimization for applications such as drug development and material discovery. Self-supervised pretraining of transformer models requires large-scale data sets, which are often sparsely populated in topical areas such as polymer science. Further, state-of-the-art approaches for polymers conduct data augmentation to generate additional samples but unavoidably incur extra computational costs. In contrast, large-scale open-source data sets are available for small molecules and provide a potential solution to data scarcity through transfer learning. In this work, we show that using transformers pretrained on small molecules and fine-tuned on polymer properties achieves comparable accuracy to those trained on augmented polymer data sets for a series of benchmark prediction tasks.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

Enhancing Adaptive Physics Refinement Simulations Through the Addition of Realistic Red Blood Cell Counts

Simulations of cancer cell transport require accurately modeling mm-scale and longer trajectories through a circulatory system containing trillions of deformable red blood cells, whose intercellular interactions require submicron fidelity. Using a hybrid CPU-GPU approach, we extend the advanced physics refinement (APR) method to couple a finely-resolved region of explicitly-modeled red blood cells to a coarsely-resolved bulk fluid domain. We further develop algorithms that: capture the dynamics at the interface of differing viscosities, maintain hematocrit within the cell-filled volume, and move the finely-resolved region and encapsulated cells while tracking an individual cancer cell. Comparison to a fully-resolved fluid-structure interaction model is presented for verification. Finally, we use the advanced APR method to simulate cancer cell transport over a mm-scale distance while maintaining a local region of RBCs, using a fraction of the computational power required to run a fully-resolved model.

Roychowdhury, Sayan↗

DOE OSTI2023

Performance Evaluation of Heterogeneous GPU Programming Frameworks for Hemodynamic Simulations

Preparing for the deployment of large scientific and engineering codes on upcoming exascale systems with GPU-dense nodes is made challenging by the unprecedented diversity of device architectures and heterogeneous programming models. In this work, we evaluate the process of porting a massively parallel, fluid dynamics code written in CUDA to SYCL, HIP, and Kokkos with a range of backends, using a combination of automated tools and manual tuning. We use a proxy application along with a custom performance model to inform the results and identify additional optimization strategies. At scale performance of the programming model implementations are evaluated on pre-production GPU node architectures for Frontier and Aurora, as well as on current NVIDIA device-based systems Summit and Polaris. Real-world workloads representing 3D blood flow calculations in complex vasculature are assessed. Our analysis highlights critical trade-offs between code performance, portability, and development time.

Martin, Aristotle↗

DOE OSTI2023

Moment Representation of Regularized Lattice Boltzmann Methods on NVIDIA and AMD GPUs

The lattice Boltzmann method is a highly scalable Navier-Stokes solver that has been applied to flow problems in a wide array of domains. However, the method is bandwidth-bound on modern GPU accelerators and has a large memory footprint. In this paper, we present new 2D and 3D GPU implementations of two different regularized lattice Boltzmann methods, which are not only able to achieve an acceleration of ∼ 1.4 × w.r.t. reference lattice Boltzmann implementations but also reduce the memory requirements by up to 35% and 47% in 2D and 3D simulations respectively. These new approaches are evaluated on NVIDIA and AMD GPU architectures.

Valero Lara, Pedro↗

DOE OSTI2023

TwoFold: Highly accurate structure and affinity prediction for protein-ligand complexes from sequences

We describe our development of ab initio protein-ligand binding pose prediction models based on transformers and binding affinity prediction models based on the neural tangent kernel (NTK). Folding both protein and ligand, the TwoFold models achieve efficient and quality predictions matching state-of-the-art implementations while additionally reconstructing protein structures. In conclusion, solving NTK models points to a new use case for highly optimized linear solver benchmarking codes on HPC.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2023

FrESCO: Framework for Exploring Scalable Computational Oncology

The National Cancer Institute (NCI) monitors population level cancer trends as part of its Surveillance, Epidemiology, and End Results (SEER) program. This program consists of state or regional level cancer registries which collect, analyze, and annotate cancer pathology reports. From these annotated pathology reports, each individual registry aggregates cancer phenotype information from electronic health records. This data is then used to create summary statistics about cancer incidence and mortality to facilitate population health monitoring. Extracting phenotypic information from these reports is a labor intensive task, requiring specialized knowledge about the reports and cancer. Automating the information extraction process from cancer pathology reports has the potential to improve data quality by extracting information in a consistent manner across registries. It can also improve patient outcomes by reducing the time from diagnosis, enabling rapid case ascertainment for clinical trials. Here we present FrESCO, a modular deep-learning natural language processing (NLP) library initially designed for extracting pathology information from clinical text documents. This repository is not solely limited to clinical medical text, but may also be used by researchers just getting started with NLP methods and those looking for a robust solution for their classification problems.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2023

Characterizing Quantum Classifier Utility in Natural Language Processing Workflows

Quantum Natural Language Processing (QNLP) develops natural language processing (NLP) models for deployment on quantum computers. We explore feature and data prototype selection techniques to address challenges posed by encoding high dimensional features. Our study builds quantum circuit classifiers that includes classical feature pre-processing, quantum embedding and quantum model training. The quantum models are built on 4 or 6 qubits and the quantum neural network (QNN) uses the established bricklayer design. We compare the dependence of model performance (in terms of accuracy and F1 scores) on feature length, embedding gates and parameterized unitary design. We compare the performance of quantum machine learning models to classical convolution neural network model (CNN) on binary and multi-class classification tasks using two datasets of synthetic features and labels. The first is the ECP-CANDLE P3B3 dataset a corpus of synthetically generated cancer pathology reports. The second dataset is extracted from well-known benchmark dataset (MADELON) - features are generated with a combination of informative, repeated and uninformative features. Both datasets are used for binary classification and multi-class classification with 3 classes. We observe robust, accurate performance from all models on the binary classification tasks, but multiclass classification is a challenge for the quantum models-there is a notable decrease in accuracy when using 3 classes. Overall the performance is comparable in terms of recall and accuracy between QNNs and CNNs, even with large datasets. These results provide a point of comparison between quantum and classical models on real-world datasets.

Hamilton, Kathleen↗

DOE OSTI2023

Evaluation of pre-training large language models on leadership-class supercomputers

Large language models (LLMs) have arisen rapidly to the center stage of artificial intelligence as the foundation models applicable to many downstream learning tasks. However, how to effectively build, train, and serve such models for many high-stake and first-principle-based scientific use cases are both of great interests and of great challenges. Moreover, pre-training LLMs with billions or even trillions of parameters can be prohibitively expensive not just for academic institutions, but also for well-funded industrial and government labs. Furthermore, the energy cost and the environmental impact of developing LLMs must be kept in mind. Here, in this work, we conduct a first-of-its-kind performance analysis to understand the time and energy cost of pre-training LLMs on the Department of Energy (DOE)’s leadership-class supercomputers. Employing state-of-the-art distributed training techniques, we evaluate the computational performance of various parallelization approaches at scale for a range of model sizes, and establish a projection model for the cost of full training. Our findings provide baseline results, best practices, and heuristics for pre-training such large models that should be valuable to HPC community at large. We also offer insights and optimization strategies for using the first exascale computing system, Frontier, to train models of the size of GPT-3 and beyond.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Adaptive language model training for molecular design

Abstract The vast size of chemical space necessitates computational approaches to automate and accelerate the design of molecular sequences to guide experimental efforts for drug discovery. Genetic algorithms provide a useful framework to incrementally generate molecules by applying mutations to known chemical structures. Recently, masked language models have been applied to automate the mutation process by leveraging large compound libraries to learn commonly occurring chemical sequences (i.e., using tokenization) and predict rearrangements (i.e., using mask prediction). Here, we consider how language models can be adapted to improve molecule generation for different optimization tasks. We use two different generation strategies for comparison, fixed and adaptive. The fixed strategy uses a pre-trained model to generate mutations; the adaptive strategy trains the language model on each new generation of molecules selected for target properties during optimization. Our results show that the adaptive strategy allows the language model to more closely fit the distribution of molecules in the population. Therefore, for enhanced fitness optimization, we suggest the use of the fixed strategy during an initial phase followed by the use of the adaptive strategy. We demonstrate the impact of adaptive training by searching for molecules that optimize both heuristic metrics, drug-likeness and synthesizability, as well as predicted protein binding affinity from a surrogate model. Our results show that the adaptive strategy provides a significant improvement in fitness optimization compared to the fixed pre-trained model, empowering the application of language models to molecular design tasks.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2023

FrESCO

The National Cancer Institute (NCI) monitors population level cancer trends as part of its Surveillance, Epidemiology, and End Results (SEER) program. This program consists of state or regional level cancer registries which collect, analyze, and annotate cancer pathology reports. From these annotated pathology reports, each individual registry aggregates cancer phenotype information and summary statistics about cancer prevalence to facilitate population level monitoring of cancer incidence. Extracting cancer phenotype from these reports is a labor intensive task, requiring specialized knowledge about the reports and cancer. Automating this information extraction process from cancer pathology reports has the potential to improve not only the quality of the data by extracting information in a consistent manner across registries, but to improve the quality of patient outcomes by reducing the time to assimilate new data and enabling time-sensitive applications such as precision medicine. Here we present FrESCO: Framework for Exploring Scalable Computational Oncology, a modular deep-learning natural language processing (NLP) library for extracting pathology information from clinical text documents.

Spannaus, Adam [Oak Ridge National Lab. (ORNL), Oa↗

DOE OSTI2023

Effect of constitutive law on the erythrocyte membrane response to large strains

Three constitutive laws, that is the Skalak, neo-Hookean and Yeoh laws, commonly employed for describing the erythrocyte membrane mechanics are theoretically analyzed and numerically investigated to assess their accuracy for capturing erythrocyte deformation characteristics and morphology. In this work, particular emphasis is given to the nonlinear deformation regime, where it is known that the discrepancies between constitutive laws are most prominent. Hence, the experiments of optical tweezers and micropipette aspiration are considered here, for which relationships between the individual shear elastic moduli of the constitutive laws can also be established through analysis of the tension-deformation relationship. All constitutive laws were found to adequately predict the axial and transverse deformations of a red blood cell subjected to stretching with optical tweezers for a constant shear elastic modulus value. As opposed to Skalak law, the neo-Hookean and Yeoh laws replicated the erythrocyte membrane folding, that has been experimentally observed, with the trade-off of sustaining significant area variations. For the micropipette aspiration, the suction pressure-aspiration length relationship could be excellently predicted for a fixed shear elastic modulus value only when Yeoh law was considered. Importantly, the neo-Hookean and Yeoh laws reproduced the membrane wrinkling at suction pressures close to those experimentally measured. None of the constitutive laws suffered from membrane area compressibility in the micropipette aspiration case.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2022

Computational Workflow for Accelerated Molecular Design Using Quantum Chemical Simulations and Deep Learning Models

Efficient methods for searching the chemical space of molecular compounds are needed to automate and accelerate the design of new functional molecules such as pharmaceuticals. Given the high cost in both resources and time for experimental efforts, computational approaches play a key role in guiding the selection of promising molecules for further investigation. Here, we construct a workflow to accelerate design by combining approximate quantum chemical methods [i.e. density-functional tight-binding (DFTB)], a graph convolutional neural network (GCNN) surrogate model for chemical property prediction, and a masked language model (MLM) for molecule generation. Property data from the DFTB calculations are used to train the surrogate model; the surrogate model is used to score candidates generated by the MLM. The surrogate reduces computation time by orders of magnitude compared to the DFTB calculations, enabling an increased search of chemical space. Furthermore, the MLM generates a diverse set of chemical modifications based on pre-training from a large compound library. We utilize the workflow to search for near-infrared photoactive molecules by minimizing the predicted HOMO-LUMO gap as the target property. Our results show that the workflow can generate optimized molecules outside of the original training set, which suggests that iterations of the workflow could be useful for searching vast chemical spaces in a wide range of design problems.

Blanchard, Andrew↗

DOE OSTI2022

Evaluation of intracoronary hemodynamics identifies perturbations in vorticity

Coronary artery disease (CAD) is highly prevalent and associated with adverse events. Challenges have emerged in the treatment of intermediate coronary artery stenoses. These lesions are often interrogated with fractional flow reserve (FFR) testing to determine if a stenosis is likely to be causative for ischemia in a cardiac territory. This invasive test requires insertion of a pressure wire into a coronary vessel. Recently computational fluid dynamics (CFD) has been used to noninvasively assess fractional flow reserve in vessels reconstructed from medical imaging data. However, many of these simulations are unable to provide additional information about intravascular hemodynamics, including velocity, endothelial shear stress (ESS), and vorticity. We hypothesized that vorticity, which has demonstrated utility in the assessment of ventricular and aortic diseases, would also be an important hemodynamic factor in CAD. Three-dimensional (3D), patient-specific coronary artery geometries that included all vessels >1 mm in diameter were created from angiography data obtained from 10 patients who underwent diagnostic angiography and FFR testing (n = 9). A massively parallel CFD solver (HARVEY) was used to calculate coronary hemodynamic parameters including pressure, velocity, ESS, and vorticity. These simulations were validated by comparing velocity flow fields from simulation to both velocities derived from in vitro particle image velocimetry and to invasively acquired pressure wire-based data from clinical testing. There was strong agreement between findings from CFD simulations and particle image velocimetry experimental testing (p < 0.01). CFD-FFR was also highly correlated with invasively measured FFR (ρ = 0.77, p = 0.01) with an average error of 5.9 ± 0.1%. CFD-FFR also had a strong inverse correlation with the vorticity (ρ = -0.86, p = 0.001). Simulations to determine the effect of the coronary stenosis on intravascular hemodynamics demonstrated significant differences in velocity and vorticity (both p < 0.05). Further evaluation of an angiographically normal appearing non-FFR coronary vessel in patients with CAD also demonstrated differences in vorticity when compared with FFR vessels (p < 0.05). The use of highly accurate 3D CFD-derived intravascular hemodynamics provides additional information beyond pressure measurements that can be used to calculate FFR. Vorticity is one parameter that is modified by a coronary stenosis and appears to be abnormal in angiographically normal vessels in patients with CAD, highlighting a possible use-case in preventative screening for early coronary disease.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2022

Language models for the prediction of SARS-CoV-2 inhibitors

The COVID-19 pandemic highlights the need for computational tools to automate and accelerate drug design for novel protein targets. We leverage deep learning language models to generate and score drug candidates based on predicted protein binding affinity. We pre-trained a deep learning language model (BERT) on ∼9.6 billion molecules and achieved peak performance of 603 petaflops in mixed precision. Our work reduces pre-training time from days to hours, compared to previous efforts with this architecture, while also increasing the dataset size by nearly an order of magnitude. For scoring, we fine-tuned the language model using an assembled set of thousands of protein targets with binding affinity data and searched for inhibitors of specific protein targets, SARS-CoV-2 Mpro and PLpro. We utilized a genetic algorithm approach for finding optimal candidates using the generation and scoring capabilities of the language model. Our generalizable models accelerate the identification of inhibitors for emerging therapeutic targets.

Blanchard, Andrew E.↗

DOE OSTI2022

High Performance Adaptive Physics Refinement to Enable Large-Scale Tracking of Cancer Cell Trajectory

The ability to track simulated cancer cells through the circulatory system, important for developing a mechanistic understanding of metastatic spread, pushes the limits of today's supercomputers by requiring the simulation of large fluid volumes at cellular-scale resolution. To overcome this challenge, we introduce a new adaptive physics refinement (APR) method that captures cellular-scale interaction across large domains and leverages a hybrid CPU-GPU approach to maximize performance. Through algorithmic advances that integrate multi-physics and multi-resolution models, we establish a finely resolved window with explicitly modeled cells coupled to a coarsely resolved bulk fluid domain. In this work we present multiple validations of the APR framework by comparing against fully resolved fluid-structure interaction methods and employ techniques, such as latency hiding and maximizing memory bandwidth, to effectively utilize heterogeneous node architectures. Collectively, these computational developments and performance optimizations provide a robust and scalable framework to enable system-level simulations of cancer cell transport.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2022

AISD HOMO-LUMO

Molecules with HOMO-LUMO gap calculated using DFTB. The molecules are a subset of a dataset generated for previous work (https://www.biorxiv.org/content/10.1101/2021.12.10.471928v1), which augmented the Enamine REAL database (https://enamine.net/). Molecules may be available for purchase through Enamine.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗