Skip to main content
eScholarship
Open Access Publications from the University of California

This series is home to publications and data sets from the Bourns College of Engineering at the University of California, Riverside.

Related Units

Center for Environmental Research and Technology

Cover page of Physics-informed UNets for discovering hidden elasticity in heterogeneous materials

Physics-informed UNets for discovering hidden elasticity in heterogeneous materials

(2024)

Soft biological tissues often have complex mechanical properties due to variation in structural components. In this paper, we develop a novel UNet-based neural network model for inversion in elasticity (El-UNet) to infer the spatial distributions of mechanical parameters from strain maps as input images, normal stress boundary conditions, and domain physics information. We show superior performance - both in terms of accuracy and computational cost - by El-UNet compared to fully-connected physics-informed neural networks in estimating unknown parameters and stress distributions for isotropic linear elasticity. We characterize different variations of El-UNet and propose a self-adaptive spatial loss weighting approach. To validate our inversion models, we performed various finite-element simulations of isotropic domains with heterogenous distributions of material parameters to generate synthetic data. El-UNet is faster and more accurate than the fully-connected physics-informed implementation in resolving the distribution of unknown fields. Among the tested models, the self-adaptive spatially weighted models had the most accurate reconstructions in equal computation times. The learned spatial weighting distribution visibly corresponded to regions that the unweighted models were resolving inaccurately. Our work demonstrates a computationally efficient inversion algorithm for elasticity imaging using convolutional neural networks and presents a potential fast framework for three-dimensional inverse elasticity problems that have proven unachievable through previously proposed methods.

Cover page of Carbon Capture: Theoretical Guidelines for Activated Carbon-Based CO2 Adsorption Material Evaluation.

Carbon Capture: Theoretical Guidelines for Activated Carbon-Based CO2 Adsorption Material Evaluation.

(2023)

Activated carbon (AC)-based materials have shown promising performance in carbon capture, offering low cost and sustainable sourcing from abundant natural resources. Despite ACs growing as a new class of materials, theoretical guidelines for evaluating their viability in carbon capture are a crucial research gap. We address this gap by developing a hierarchical guideline, based on fundamental gas-solid interaction strength, that underpins the success and scalability of AC-based materials. The most critical performance indicator is the CO2 adsorption energy, where an optimal range (-0.41 eV) ensures efficiency between adsorption and desorption. Additionally, we consider thermal stability and defect sensitivity to ensure consistent performance under varying conditions. Further, selectivity and capacity play significant roles due to external variables such as partial pressure of CO2 and other ambient air gases (N2, H2O, O2), bridging the gap between theory and reality. We provide actionable examples by narrowing our options to methylamine- and pyridine-grafted graphene.

Cover page of Harnessing Semi-Supervised Machine Learning to Automatically Predict Bioactivities of Per- and Polyfluoroalkyl Substances (PFASs).

Harnessing Semi-Supervised Machine Learning to Automatically Predict Bioactivities of Per- and Polyfluoroalkyl Substances (PFASs).

(2023)

Many per- and polyfluoroalkyl substances (PFASs) pose significant health hazards due to their bioactive and persistent bioaccumulative properties. However, assessing the bioactivities of PFASs is both time-consuming and costly due to the sheer number and expense of in vivo and in vitro biological experiments. To this end, we harnessed new unsupervised/semi-supervised machine learning models to automatically predict bioactivities of PFASs in various human biological targets, including enzymes, genes, proteins, and cell lines. Our semi-supervised metric learning models were used to predict the bioactivity of PFASs found in the recent Organisation of Economic Co-operation and Development (OECD) report list, which contains 4730 PFASs used in a broad range of industries and consumers. Our work provides the first semi-supervised machine learning study of structure-activity relationships for predicting possible bioactivities in a variety of PFAS species.

Cover page of Enhancing the Combustion of Magnesium Nanoparticles via Low-Temperature Plasma-Induced Hydrogenation.

Enhancing the Combustion of Magnesium Nanoparticles via Low-Temperature Plasma-Induced Hydrogenation.

(2023)

The hydrogenation of metal nanoparticles provides a pathway toward tuning their combustion characteristics. Metal hydrides have been employed as solid-fuel additives for rocket propellants, pyrotechnics, and explosives. Gas generation during combustion is beneficial to prevent aggregation and sintering of particles, enabling a more complete fuel utilization. Here, we discuss a novel approach for the synthesis of magnesium hydride nanoparticles based on a two-step aerosol process. Mg particles are first nucleated and grown via thermal evaporation, followed immediately by in-flight exposure to a hydrogen-rich low-temperature plasma. During the second step, atomic hydrogen generated by the plasma rapidly diffuses into the Mg lattice, forming particles with a significant fraction of MgH2. We find that hydrogenated Mg nanoparticles have an ignition temperature that is reduced by ∼200 °C when combusted with potassium perchlorate as an oxidizer, compared to the non-hydrogenated Mg material. This is due to the release of hydrogen from the fuel, jumpstarting its combustion. In addition, characterization of the plasma processes suggests that a careful balance between the dissociation of molecular hydrogen and heating of the nanoparticles must be achieved to avoid hydrogen desorption during production and achieve a significant degree of hydrogenation.

Cover page of Prioritised identification of structural classes of natural products from higher plants in the expedition of antimalarial drug discovery.

Prioritised identification of structural classes of natural products from higher plants in the expedition of antimalarial drug discovery.

(2023)

The emergence and spread of drug-recalcitrant Plasmodium falciparum parasites threaten to reverse the gains made in the fight against malaria. Urgent measures need to be taken to curb this impending challenge. The higher plant-derived sesquiterpene, quinoline alkaloids, and naphthoquinone natural product classes of compounds have previously served as phenomenal chemical scaffolds from which integral antimalarial drugs were developed. Historical successes serve as an inspiration for the continued investigation of plant-derived natural products compounds in search of novel molecular templates from which new antimalarial drugs could be developed. The aim of this study was to identify potential chemical scaffolds for malaria drug discovery following analysis of historical data on phytochemicals screened in vitro against P. falciparum. To identify these novel scaffolds, we queried an in-house manually curated database of plant-derived natural product compounds and their in vitro biological data. Natural products were assigned to different structural classes using NPClassifier. To identify the most promising chemical scaffolds, we then correlated natural compound class with bioactivity and other data, namely (i) potency, (ii) resistance index, (iii) selectivity index and (iv) physicochemical properties. We used an unbiased scoring system to rank the different natural product classes based on the assessment of their bioactivity data. From this analysis we identified the top-ranked natural product pathway as the alkaloids. The top three ranked super classes identified were (i) pseudoalkaloids, (ii) naphthalenes and (iii) tyrosine alkaloids and the top five ranked classes (i) quassinoids (of super class triterpenoids), (ii) steroidal alkaloids (of super class pseudoalkaloids) (iii) cycloeudesmane sesquiterpenoids (of super class triterpenoids) (iv) isoquinoline alkaloids (of super class tyrosine alkaloids) and (v) naphthoquinones (of super class naphthalenes). Launched chemical space of these identified classes of compounds was, by and large, distinct from that of legacy antimalarial drugs. Our study was able to identify chemical scaffolds with acceptable biological properties that are structurally different from current and previously used antimalarial drugs. These molecules have the potential to be developed into new antimalarial drugs.

Cover page of Leveraging off higher plant phylogenetic insights for antiplasmodial drug discovery.

Leveraging off higher plant phylogenetic insights for antiplasmodial drug discovery.

(2023)

The antimalarial drug-resistance conundrum which threatens to reverse the great strides taken to curb the malaria scourge warrants an urgent need to find novel chemical scaffolds to serve as templates for the development of new antimalarial drugs. Plants represent a viable alternative source for the discovery of unique potential antiplasmodial chemical scaffolds. To expedite the discovery of new antiplasmodial compounds from plants, the aim of this study was to use phylogenetic analysis to identify higher plant orders and families that can be rationally prioritised for antimalarial drug discovery. We queried the PubMed database for publications documenting antiplasmodial properties of natural compounds isolated from higher plants. Thereafter, we manually collated compounds reported along with plant species of origin and relevant pharmacological data. We systematically assigned antiplasmodial-associated plant species into recognised families and orders, and then computed the resistance index, selectivity index and physicochemical properties of the compounds from each taxonomic group. Correlating the generated phylogenetic trees and the biological data of each clade allowed for the identification of 3 hot plant orders and families. The top 3 ranked plant orders were the (i) Caryophyllales, (ii) Buxales, and (iii) Chloranthales. The top 3 ranked plant families were the (i) Ancistrocladaceae, (ii) Simaroubaceae, and (iii) Buxaceae. The highly active natural compounds (IC50 ≤ 1 µM) isolated from these plant orders and families are structurally unique to the legacy antimalarial drugs. Our study was able to identify the most prolific taxa at order and family rank that we propose be prioritised in the search for potent, safe and drug-like antimalarial molecules.

Cover page of Babesia BdFE1 Esterase is Required for the Anti-parasitic Activity of the ACE Inhibitor Fosinopril

Babesia BdFE1 Esterase is Required for the Anti-parasitic Activity of the ACE Inhibitor Fosinopril

(2023)

Effective and safe therapies for the treatment of diseases caused by intraerythrocytic parasites are impeded by the rapid emergence of drug resistance and the lack of novel drug targets. One such disease is human babesiosis, which is a rapidly emerging tick-borne illness caused by Babesia parasites. In this study, we identified fosinopril, a phosphonate-containing, FDA-approved Angiotensin Converting Enzyme (ACE) inhibitor commonly used as a prodrug for hypertension and heart failure, as a potent inhibitor of B. duncani parasite development within human erythrocytes. Cell biological and mass spectrometry analyses revealed that the conversion of fosinopril to its active diacid molecule, fosinoprilat, is essential for its antiparasitic activity. We show that this conversion is mediated by a parasite-encoded esterase, BdFE1, which is highly conserved among apicomplexan parasites. Parasites carrying the L238H mutation in the active site of BdFE1 failed to convert the prodrug to its active moiety and became resistant to the drug. Our data set the stage for the development of this class of drugs for therapy of vector-borne parasitic diseases.

Cover page of FLUID-GPT (Fast Learning to Understand and Investigate Dynamics with a Generative Pre-Trained Transformer): Efficient Predictions of Particle Trajectories and Erosion.

FLUID-GPT (Fast Learning to Understand and Investigate Dynamics with a Generative Pre-Trained Transformer): Efficient Predictions of Particle Trajectories and Erosion.

(2023)

The deleterious impact of erosion due to high-velocity particle impingement adversely affects a variety of engineering and industrial systems, resulting in irreversible mechanical wear of materials/components. Brute force computational fluid dynamics (CFD) calculations are commonly used to predict surface erosion by directly solving the Navier-Stokes equations for fluid and particle dynamics; however, these numerical approaches often require significant computational resources. In contrast, recent data-driven approaches using machine learning (ML) have shown immense promise for more efficient and accurate predictions to sidestep computationally demanding CFD calculations. To this end, we have developed FLUID-GPT (Fast Learning to Understand and Investigate Dynamics with a Generative Pre-Trained Transformer), a new hybrid ML architecture for accurately predicting particle trajectories and erosion on an industrial-scale steam header geometry. Our FLUID-GPT approach utilizes a Generative Pre-Trained Transformer 2 (GPT-2) with a convolutional neural network (CNN) for the first time to predict surface erosion using only information from five initial conditions: particle size, main-inlet speed, main-inlet pressure, subinlet speed, and subinlet pressure. Compared to the bidirectional long- and short-term memory (BiLSTM) ML techniques used in previous work, our FLUID-GPT model is much more accurate (a 54% decrease in the mean squared error) and efficient (70% less training time). Our work demonstrates that FLUID-GPT is an accurate and efficient ML approach for predicting time-series trajectories and their subsequent spatial erosion patterns in these complex dynamic systems.

Cover page of Indocyanine Green (ICG) Fluorescence Is Dependent on Monomer with Planar and Twisted Structures and Inhibited by H-Aggregation

Indocyanine Green (ICG) Fluorescence Is Dependent on Monomer with Planar and Twisted Structures and Inhibited by H-Aggregation

(2023)

The optical properties of indocyanine green (ICG) as a near-infrared (NIR) fluorescence dye depend on the nature of the solvent medium and the dye concentration. In the ICG absorption spectra of water, at high concentrations, there were absorption maxima at 700 nm, implying H-aggregates. With ICG dilution, the main absorption peak was at 780 nm, implying monomers. However, in ethanol, the absorption maximum was 780 nm, and the shapes of the absorption spectra were identical regardless of the ICG concentration, indicating that ICG in ethanol exists only as a monomer without H-aggregates. We found that emission was due to the monomer form and decreased with H-aggregate formation. In the fluorescence spectra, the 820 nm emission band was dominant at low concentrations, whereas at high concentrations, we found that the emission peaks were converted to 880 nm, suggesting a new form via the twisted intramolecular charge transfer (TICT) process of ICG. The NIR fluorescence intensity of ICG in ethanol was approximately 12- and 9-times brighter than in water in the NIR-I and -II regions, respectively. We propose an energy diagram of ICG to describe absorptive and emissive transitions through the ICG structures such as the monomer, H-aggregated, and TICT monomer forms.