Engineering topics

Xu, Weihong

Publications and source records attributed to Xu, Weihong.

4 records

DOE OSTI2024

HDBind: encoding of molecular structure with hyperdimensional binary representations

Traditional methods for identifying “hit” molecules from a large collection of potential drug-like candidates rely on biophysical theory to compute approximations to the Gibbs free energy of the binding interaction between the drug and its protein target. These approaches have a significant limitation in that they require exceptional computing capabilities for even relatively small collections of molecules. Increasingly large and complex state-of-the-art deep learning approaches have gained popularity with the promise to improve the productivity of drug design, notorious for its numerous failures. However, as deep learning models increase in their size and complexity, their acceleration at the hardware level becomes more challenging. Hyperdimensional Computing (HDC) has recently gained attention in the computer hardware community due to its algorithmic simplicity relative to deep learning approaches. The HDC learning paradigm, which represents data with high-dimension binary vectors, allows the use of low-precision binary vector arithmetic to create models of the data that can be learned without the need for the gradient-based optimization required in many conventional machine learning and deep learning methods. This algorithmic simplicity allows for acceleration in hardware that has been previously demonstrated in a range of application areas (computer vision, bioinformatics, mass spectrometery, remote sensing, edge devices, etc.). To the best of our knowledge, our work is the first to consider HDC for the task of fast and efficient screening of modern drug-like compound libraries. We also propose the first HDC graph-based encoding methods for molecular data, demonstrating consistent and substantial improvement over previous work. We compare our approaches to alternative approaches on the well-studied MoleculeNet dataset and the recently proposed LIT-PCBA dataset derived from high quality PubChem assays. We demonstrate our methods on multiple target hardware platforms, including Graphics Processing Units (GPUs) and Field Programmable Gate Arrays (FPGAs), showing at least an order of magnitude improvement in energy efficiency versus even our smallest neural network baseline model with a single hidden layer. Our work thus motivates further investigation into molecular representation learning to develop ultra-efficient pre-screening tools. We make our code publicly available at https://github.com/LLNL/hdbind.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2023

HD-Bind: Encoding of Molecular Structure with Low Precision, Hyperdimensional Binary Representations

Publicly available collections of drug-like molecules have grown to comprise tens of billions of compounds due to advances in combinatorial chemistry. Traditional methods for identifying "hit" molecules from a large collection of potential drug-like candidates have relied on biophysical theory to compute approximations to the Gibbs free energy of the binding interaction between the drug and its protein target. These approaches have the major drawback that they require exceptional computing capabilities for even relatively small collections of molecules. Hyperdimensional Computing (HDC) is a recently-proposed learning paradigm that represents data with high-dimension binary vectors; this allows the use of low-precision binary vector arithmetic to create models of the data that can be learned without the need for the gradient-based optimization required in many conventional machine learning and deep learning methods. This algorithmic simplicity allows for acceleration in hardware that has been previously demonstrated in a range of application areas. We consider existing HDC approaches for molecular property classification and introduce two novel encodings of a commonly-used molecular representation, the extended connectivity fingerprint (ECFP). We show that HDC-based inference methods are as much as 91 times more efficient than traditional machine learning methods, and achieve an acceleration of nearly nine orders of magnitude compared to molecular docking. Our results show that HDC accelerated methods retain competitive accuracy on a number of well-studied tasks such as molecular property predictions using the MoleculeNet dataset, and bind/no-bind activity classification using the DUD-E and LIT-PCBA datasets. Our work thus motivates further investigation into molecular representation learning to develop ultraefficient pre-screening tools.

Jones, William↗

DOE OSTI2015

The novel HDAC inhibitor AR-42-induced anti-colon cancer cell activity is associated with ceramide production

In the current study, we investigated the potential activity of AR-42, a novel histone deacetylase (HDAC) inhibitor, against colon cancer cells. Our in vitro results showed that AR-42 induced ceramide production, exerted potent anti-proliferative and pro-apoptotic activities in established (SW-620 and HCT-116 lines) and primary human colon cancer cells. Exogenously-added sphingosine 1-phosphate (S1P) suppressed AR-42-induced activity, yet a cell-permeable ceramide (C4) facilitated AR-42-induced cytotoxicity against colon cancer cells. In addition, AR-42-induced ceramide production and anti-colon cancer cell activity were inhibited by the ceramide synthase inhibitor fumonisin B1, but were exacerbated by PDMP, which is a ceramide glucosylation inhibitor. In vivo, oral administration of a single dose of AR-42 dramatically inhibited SW-620 xenograft growth in severe combined immunodeficient (SCID) mice, without inducing overt toxicities. Together, these results show that AR-42 dramatically inhibits colon cancer cell proliferation in vitro and in vivo, and ceramide production might be the key mechanism responsible for its actions. - Highlights: • AR-42 is anti-proliferative against primary/established colon cancer cells. • AR-42 induces significant apoptotic death in primary/established colon cancer cells. • Ceramide production mediates AR-42-induced cytotoxicity in colon cancer cells. • AR-42 oral administration potently inhibits SW-620 xenograft growth in SCID mice.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2013

Genomic responses in mouse models poorly mimic human inflammatory diseases

Murine models have been extensively used in recent decades to identify and test drug candidates for subsequent human trials (1–3). However, few of these human trials have shown success (4–7). The success rate is even worse for those trials in the field of inflammation, a condition present in many human diseases. To date, there have been nearly 150 clinical trials testing candidate agents intended to block the inflammatory response in critically ill patients, and every one of these trials failed (8–11). Despite commentaries that question the merit of an overreliance of animal systems to model human immunology (3, 12, 13), in the absence of systematic evidence, investigators and public regulators assume that results from animal research reflect human disease. To date, there have been no studies to systematically evaluate, on a molecular basis, how well the murine clinical models mimic human inflammatory diseases in patients. The Inflammation and Host Response to Injury, Large Scale Collaborative Research Program has completed multiple studies on the genomic responses to systemic inflammation in patients and human volunteers as well as murine models (14–18). These datasets include genome-wide expression analysis on white blood cells obtained from serial blood draws in 167 patients up to 28 d after severe blunt trauma (15), 244 patients up to 1 y after burn injury, and 4 healthy humans for 24 h after administration of low-dose bacterial endotoxin (14) and expression analysis on analogous samples from well-established mouse models of trauma, burns, and endotoxemia (16 treated and 16 controls per model) (16–18). In humans, severe inflammatory stress produces a genomic storm affecting all major cellular functions and pathways (15) and therefore, provided sufficient perturbations to allow comparisons between the genes in the human conditions and their orthologs in the murine models. In this article, we report on a systematic comparison of the genomic response between human inflammatory diseases andmurine models. First, we compared the correlations of gene expression changes with trauma, burns, and endotoxemia between human subjects and corresponding mouse models. Second, we characterized and compared the temporal gene response patterns seen in these human conditions and models. Third, we also identified the major signaling pathways significantly regulated in the inflammatory response to human injuries and compared them with the human in vivo endotoxemia model and three murine models. Fourth, we sought and evaluated representative patient and murine studies of several additional acute inflammatory diseases. These results show that the genomic responses to different acute inflammatory stresses are highly similar in humans, but these responses are not reproduced in the current mouse models. New approaches need be explored to improve the ways that human diseases are studied.

Seok, Junhee↗