Computational Scheme Research Articles

Minimizing Carbon Emissions by Improving Water and Energy Use Efficiencies in AI Servers: A Green Cloud Computing Strategy for Sustainable Artificial Intelligence Systems

The advent of Artificial Intelligence systems, in particular of generative models like ChatGPT, has resulted in one more area requiring heavy computational resources which in turn consumes a lot of energy and water. By estimations, one interaction with ChatGPT for instance will take an estimate of 2.9 watt hour, which is ten times higher than the amount of energy consumed to conduct an ordinary googling task that is 0.3 watt hours. This stands as a call for action toward improving the water to energy ratio of the AI systems and therefore the recent carbon emissions. This paper explores the energy efficiency patterns of AI languages such as chatbots compared with the other means of searching the internet like Google and how the effects of the AI machines on the environment can be reduced. In this connection, green cloud computing methods have been suggested as possible solutions that can be effectively combined with the principles of clean energy use; on this list are both advanced systems for maintaining low temperatures and the optimization of AI systems. Finally, using of resources could also play a crucial part in the ultimate decrease in the adverse effects that the AI industry has on our environment.

Enhancing protein aggregation prediction: a unified analysis leveraging graph convolutional networks and active learning.

Protein aggregation (PA) is a critical phenomenon associated with Alzheimer's and Parkinson's disease. Recent studies have suggested that factors like aggregation-prone regions (APRs) and β-strand interactions are crucial in understanding such behavior. While experimental methods have provided valuable insights, there has been a shift towards computational strategies, particularly machine learning, for their efficacy and speed. The challenge, however, lies in effectively incorporating structural information into these models. This study constructs a Graph Convolutional Network (GCN) to predict PA scores with the expanded and refined Protein Data Bank (PDB) and AlphaFold2.0 dataset. We employed AGGRESCAN3D 2.0 to calculate PA propensity and to enhance the dataset, we systematically separated multi polypeptide chains within PDB data into single polypeptide chains, removing redundancy. This effort resulted in a dataset comprising 302 032 unique PDB entries. Subsequently, we compared sequence similarity and obtained 22 774 Homo sapiens data from AlphaFold2.0. Using this expanded and refined dataset, the trained GCN model for PA prediction achieves a remarkable coefficient of determination (R 2) score of 0.9849 and a low mean absolute error (MAE) of 0.0381. Furthermore, the efficacy of the active learning process was demonstrated through its rapid identification of proteins with high PA propensity. Consequently, the active learning approach achieved an MAE of 0.0291 in expected improvement, surpassing other methods. It identified 99% of the target proteins by exploring merely 29% of the entire search space. This improved GCN model demonstrates promise in selecting proteins susceptible to PA, advancing protein science. This work contributes to the development of efficient computational tools for PA prediction, with potential applications in disease diagnosis and therapy.

Quelling the Geometry Factor Effect in Quantum Chemical Calculations of 13C NMR Chemical Shifts with the Aid of the pecG-n (n = 1, 2) Basis Sets

A root factor for the accuracy of all quantum chemical calculations of nuclear magnetic resonance (NMR) chemical shifts is the quality of the molecular equilibrium geometry used. In turn, this quality depends largely on the basis set employed at the geometry optimization stage. This parameter represents the main subject of the present study, which is a continuation of our recent work, where new pecG-n (n = 1, 2) basis sets for the geometry optimization were introduced. A goal of this study was to compare the performance of our geometry-oriented pecG-n (n = 1, 2) basis sets against the other basis sets in massive calculations of 13C NMR shielding constants/chemical shifts in terms of their efficacy in reducing geometry factor errors. The testing was carried out with both large-sized biologically active natural products and medium-sized compounds with complicated electronic structures. The former were treated using the computation protocol based on the density functional theory (DFT) and considered in the theoretical benchmarking, while the latter were treated using the computational scheme based on the upper-hierarchy coupled cluster (CC) methods and were used in the practical benchmarking involving the comparison with experimental NMR data. Both the theoretical and practical analyses showed that the pecG-1 and pecG-2 basis sets resulted in substantially reduced geometry factor errors in the calculated 13C NMR chemical shifts/shielding constants compared to their commensurate analogs, with the pecG-2 basis set being the best of all the considered basis sets.

A Spatial-Channel-Temporal-Fused Attention for Spiking Neural Networks.

Spiking neural networks (SNNs) mimic brain computational strategies, and exhibit substantial capabilities in spatiotemporal information processing. As an essential factor for human perception, visual attention refers to the dynamic process for selecting salient regions in biological vision systems. Although visual attention mechanisms have achieved great success in computer vision applications, they are rarely introduced into SNNs. Inspired by experimental observations on predictive attentional remapping, we propose a new spatial-channel-temporal-fused attention (SCTFA) module that can guide SNNs to efficiently capture underlying target regions by utilizing accumulated historical spatial-channel information in the present study. Through a systematic evaluation on three event stream datasets (DVS Gesture, SL-Animals-DVS, and MNIST-DVS), we demonstrate that the SNN with the SCTFA module (SCTFA-SNN) not only significantly outperforms the baseline SNN (BL-SNN) and two other SNN models with degenerated attention modules, but also achieves competitive accuracy with the existing state-of-the-art (SOTA) methods. Additionally, our detailed analysis shows that the proposed SCTFA-SNN model has strong robustness to noise and outstanding stability when faced with incomplete data, while maintaining acceptable complexity and efficiency. Overall, these findings indicate that incorporating appropriate cognitive mechanisms of the brain may provide a promising approach to elevate the capabilities of SNNs.

Investigation of Surge Phenomena in a Centrifugal Compressor by 3D-1D Coupling Approach

Abstract Critical instability, such as surge phenomena, can occur at low flow rates for a compressor system, leading to flow rate fluctuations and potential damage. To study this, traditional numerical analysis requires a comprehensive approach encompassing the compressor and its piping system. This demands extensive computational resources and time due to the complexity of capturing internal flow during surge, including separation and vortex generation. To overcome these limitations, this research proposes a computational strategy combining 3D and 1D analysis for enhanced efficiency and reduced computational demands, which employs 1D analysis for more straightforward flow regions like piping and 3D analysis for the more complex compressor section. This approach is facilitated using AMESim for 1D and STAR-CCM+ for 3D simulations, enabling seamless integration and data exchange between different dimensional analyses. The method improves computational efficiency and reasonably approximates surge phenomena compared to experimental data. The 3D and 1D coupled analysis closely aligns with experimental data, providing a viable and more effective alternative method for surge analysis in compressor systems. This methodology promises to streamline the study of compressor surge phenomena, providing a practical tool for engineers and researchers.

494P Evaluation of transcriptome forward computational strategies to improve molecular diagnosis in rare pediatric neuromuscular disease

494P Evaluation of transcriptome forward computational strategies to improve molecular diagnosis in rare pediatric neuromuscular disease

Buckets Instead of Umbrellas for Enhanced Sampling and Free Energy Calculations.

Umbrella sampling has been a workhorse for free energy calculations in molecular simulations for several decades. In conventional umbrella sampling, restraining bias potentials are strategically applied along one or several collective variables. Major drawbacks associated with this method are the requirement of a large number of bias windows and the poor sampling of the transverse coordinates. In this work, we propose an alternate formalism that departs from the traditional umbrella sampling to mitigate these issues, where we replace umbrella-type restraining bias potentials with bucket-type wall potentials. This modification permits one to formulate an efficient computational strategy leveraging wall potentials and metadynamics sampling. This new method, called "bucket sampling", can significantly reduce the computational cost of obtaining converged high-dimensional free energy surfaces. Extensions of the proposed method with temperature acceleration and replica exchange solute tempering are also demonstrated.

An accurate wavelets‐collocation technique for neutral delay distributed‐order fractional optimal control problems

AbstractIn this study, a new class of optimal control problems called neutral delay distributed‐order fractional optimal control problems is introduced, this problem is solved based on an efficient computational scheme. To solve the problem, we derive an exact formula for the Riemann–Liouville fractional integral operator of Genocchi wavelets based on beta functions for the first time. By taking into account this operator, collocation method, and Gauss–Legendre integration formula, the solution of fractional optimal control problems (FOCPs) under consideration is converted to a nonlinear programming one to which existing well‐developed algorithms may be applied. The mentioned scheme is applied to both FOCPs with or without delay. Error analysis associated with the proposed idea is also investigated under several mild conditions. The effectiveness of the strategy is showed by several illustrative examples, furthermore, a comparison with the previous methods highlights the preference of this scheme.

Traversing chemical space with active deep learning for low-data drug discovery.

Deep learning is accelerating drug discovery. However, current approaches are often affected by limitations in the available data, in terms of either size or molecular diversity. Active deep learning has high potential for low-data drug discovery, as it allows iterative model improvement during the screening process. However, there are several 'known unknowns' that limit the wider adoption of active deep learning in drug discovery: (1) what the best computational strategies are for chemical space exploration, (2) how active learning holds up to traditional, non-iterative, approaches and (3) how it should be used in the low-data scenarios typical of drug discovery. To provide answers, this study simulates a low-data drug discovery scenario, and systematically analyzes six active learning strategies combined with two deep learning architectures, on three large-scale molecular libraries. We identify the most important determinants of success in low-data regimes and show that active learning can achieve up to a sixfold improvement in hit discovery when compared with traditional screening methods.

Intermolecular Bending States and Tunneling Splittings of Water Trimer from Rigorous 9D Quantum Calculations: I. Methodology, Energy Levels, and Low-Frequency Spectrum.

We present the computational methodology that enables the first rigorous nine-dimensional (9D) quantum calculations of the intermolecular bending states of the water trimer, as well as its low-frequency spectrum for direct comparison with experiment. The water monomers, treated as rigid, have their centers of mass (cm's) at the corners of an equilateral triangle, and the intermonomer cm-to-cm distance is set to a value slightly larger than that in the equilibrium geometry of the trimer. The remaining nine strongly coupled large-amplitude bending (angular) degrees of freedom (DOFs) enter the 9D bend Hamiltonian of the three coupled 3D rigid-water hindered rotors. Its 9D eigenstates encompass excited librational vibrations of the trimer, as well as their torsional and bifurcation tunneling splittings, which have been the subject of much interest. The calculations of these eigenstates are extremely demanding, and a sophisticated computational scheme is developed that exploits the molecular symmetry group of the water trimer, G48, in order to make them feasible in a reasonable amount of time. The spectrum of the low-frequency vibrations of the water trimer simulated using the eigenstates of the 9D bend Hamiltonian agrees remarkably well with the experimentally observed far-infrared (FIR) spectrum of the trimer in helium nanodroplets over the entire frequency range of the measurements from 70 to 620 cm-1. This shows that most peaks in the experimental FIR spectrum are associated with the intermolecular bending vibrations of the trimer. Moreover, the ground-state torsional tunneling splittings from the present 9D calculations are in excellent agreement with the spectroscopic data. These results demonstrate the high quality of the ab initio 2 + 3-body PES employed for the DOFs included in the bound-state calculations.

Hamilton-Jacobi-Bellman equations for Rydberg-blockade processes

We discuss time-optimal control problems for two setups involving globally driven Rydberg atoms in the blockade limit by deriving the associated Hamilton-Jacobi-Bellman equations. From these equations, we extract the globally optimal trajectories and the corresponding controls for several target processes of the atomic system, using a generalized method of characteristics. We apply this method to retrieve known results for CZ and C-phase gates, and to find new optimal pulses for all elementary processes involved in the universal quantum computation scheme introduced in []. Published by the American Physical Society 2024

Generative AI in orthopedics: an explainable deep few-shot image augmentation pipeline for plain knee radiographs and Kellgren-Lawrence grading.

Recently, deep learning medical image analysis in orthopedics has become highly active. However, progress has been restricted by the absence of large-scale and standardized ground-truth images. To the best of our knowledge, this study is the first to propose an innovative solution, namely a deep few-shot image augmentation pipeline, that addresses this challenge by synthetically generating knee radiographs for training downstream tasks, with a specific focus on knee osteoarthritis Kellgren-Lawrence (KL) grading. This study leverages a deep few-shot image augmentation pipeline to generate synthetic knee radiographs. Despite the limited availability of training samples, we demonstrate the capability of our proposed computational strategy to produce high-fidelity plain knee radiographs and use them to successfully train a KL grade classifier. Our experimental results showcase the effectiveness of the proposed computational pipeline. The generated synthetic radiographs exhibit remarkable fidelity, evidenced by the achieved average Frechet Inception Distance (FID) score of 26.33 for KL grading and 22.538 for bilateral knee radiographs. For KL grading classification, the classifier achieved a test Cohen's Kappa and accuracy of 0.451 and 0.727, respectively. Our computational strategy also resulted in a publicly and freely available imaging dataset of 86000 synthetic knee radiographs. Our approach demonstrates the capability to produce top-notch synthetic knee radiographs and use them for KL grading classification, even when working with a constrained training dataset. The results obtained emphasize the effectiveness of the pipeline in augmenting datasets for knee osteoarthritis research, opening doors for broader applications in orthopedics, medical image analysis, and AI-powered diagnosis.

Revamped role for approved drug: integrative computational and biophysical analysis of saquinavir's peptidyl arginine deiminase 4 inhibition for rheumatoid arthritis.

The pursuit of novel therapeutics is a complex and resource-intensive endeavor marked by significant challenges, including high costs and low success rates. In response, drug repositioning strategies leverage existing FDA-approved compounds to predict their efficacy across diverse diseases. Peptidyl arginine deiminase 4 (PAD4) plays a pivotal role in protein citrullination, a process implicated in the autoimmune pathogenesis of rheumatoid arthritis (RA). Targeting PAD4 has thus emerged as a promising therapeutic approach. This study employs computational and enzyme inhibition strategies to identify potential PAD4-targeting compounds from a library of FDA-approved drugs. In silico docking analyses validated the binding interactions and orientations of screened compounds within PAD4's active site, with key residues such as ASP350, HIS471, ASP473, and CYS645 participating in crucial hydrogen bonding and van der Waals interactions. Molecular dynamics simulations further assessed the stability of top compounds exhibiting high binding affinities. Among these compounds, Saquinavir (SQV) emerged as a potent PAD4 inhibitor, demonstrating competitive inhibition with a low IC50 value of 1.21 ± 0.04 µM. In vitro assays, including enzyme kinetics and biophysical analyses, highlighted significant changes in PAD4 conformation upon SQV binding, as confirmed by circular dichroism spectroscopy. SQV induced localized alterations in PAD4 structure, effectively occupying the catalytic pocket and inhibiting enzymatic activity. These findings underscore SQV's potential as a therapeutic candidate for RA through PAD4 inhibition. Further validation through in vitro and in vivo studies is essential to confirm SQV's therapeutic benefits in autoimmune diseases associated with dysregulated citrullination.

SiRNADiscovery: a graph neural network for siRNA efficacy prediction via deep RNA sequence analysis.

The clinical adoption of small interfering RNAs (siRNAs) has prompted the development of various computational strategies for siRNA design, from traditional data analysis to advanced machine learning techniques. However, previous studies have inadequately considered the full complexity of the siRNA silencing mechanism, neglecting critical elements such as siRNA positioning on mRNA, RNA base-pairing probabilities, and RNA-AGO2 interactions, thereby limiting the insight and accuracy of existing models. Here, we introduce siRNADiscovery, a Graph Neural Network (GNN) framework that leverages both non-empirical and empirical rule-based features of siRNA and mRNA to effectively capture the complex dynamics of gene silencing. On multiple internal datasets, siRNADiscovery achieves state-of-the-art performance. Significantly, siRNADiscovery also outperforms existing methodologies in in vitro studies and on an externally validated dataset. Additionally, we develop a new data-splitting methodology that addresses the data leakage issue, a frequently overlooked problem in previous studies, ensuring the robustness and stability of our model under various experimental settings. Through rigorous testing, siRNADiscovery has demonstrated remarkable predictive accuracy and robustness, making significant contributions to the field of gene silencing. Furthermore, our approach to redefining data-splitting standards aims to set new benchmarks for future research in the domain of predictive biological modeling for siRNA.

Proven Distributed Memory Parallelization of Particle Methods

We provide a mathematically proven parallelization scheme for particle methods on distributed-memory computer systems. Particle methods are a versatile and widely used class of algorithms for computer simulations and numerical predictions in various applications, ranging from continuum fluid dynamics to discrete granular flows and molecular dynamics simulations. Particle methods naturally lend themselves to implementation on parallel computing systems. So far, however, a mathematical proof of correctness and equivalence to sequential implementations was only available for shared-memory parallelism. Here, we leverage a formal definition of the algorithmic class of particle methods to provide a proven parallelization scheme for distributed-memory computers. We prove that thus parallelized particle methods on distributed-memory computers are formally equivalent to their sequential counterpart for a well-defined class of particle methods, and we provide analytical expressions for the speed-up and scalability bounds of this class of algorithms in function of their parameters. The parallelization scheme analyzed here is the basis of many real-world software designs for parallel particle methods. The present analysis is, therefore, of direct relevance to existing and new parallel implementations of particle methods and places them on solid theoretical grounds, rationalizing best practices and providing useful scalability and speedup bounds for benchmarking.

Measurement of adhesion and traction of cells at high yield reveals an energetic ratchet operating during nephron condensation

Developmental biology-inspired strategies for tissue-building have extraordinary promise for regenerative medicine, spurring interest in the relationship between cell biophysical properties and morphological transitions. However, mapping gene or protein expression data to cell biophysical properties to physical morphogenesis remains challenging with current techniques. Here, we present multiplexed adhesion and traction of cells at high yield (MATCHY). MATCHY advances the multiplexing and throughput capabilities of existing traction force and cell-cell adhesion assays using microfabrication and a semiautomated computation scheme with machine learning-driven cell segmentation. Both biophysical assays are coupled with serial downstream immunofluorescence to extract cell type/signaling state information. MATCHY is especially suited to complex primary tissue-, organoid-, or biopsy-derived cell mixtures since it does not rely on a priori knowledge of cell surface markers, cell sorting, or use of lineage-specific reporter animals. We first validate MATCHY on canine kidney epithelial cells engineered for rearranged during transfection (RET) tyrosine kinase expression and quantify a relationship between downstream signaling and cell traction. We then use MATCHY to create a biophysical atlas of mouse embryonic kidney primary cells and identify distinct biophysical states along the nephron differentiation trajectory. Our data complement expression-level knowledge of adhesion molecule changes that accompany nephron differentiation with quantitative biophysical information. These data reveal an "energetic ratchet" that accounts for spatial trends in nephron progenitor cell condensation as they differentiate into early nephron structures, which we validate through agent-based computational simulation. MATCHY offers semiautomated cell biophysical characterization at >10,000-cell throughput, an advance benefiting fundamental studies and new synthetic tissue strategies for regenerative medicine.

Acoustic emission and multiscale computation-guided tensile damage identification in woven composite laminates at cryogenic temperatures as low as 20 K

For laminated composite structures as key components of storage tanks serving at cryogenic temperature, it is crucial to identify the damage mechanisms for evaluating their mechanical properties and guiding structural design. In this work, the cryogenic tensile damage behavior of a thin-walled woven composite laminate was investigated through the acoustic emission (AE) monitoring and multiscale finite element (FE) computation at typical low temperatures of 153 K, 77 K and 20 K. We first established temperature-dependent constitutive laws for the microscale and mesoscale constituents of such composites based on experimental data, followed by the development of a hierarchical computational framework for modeling multiscale damage characteristics at different low temperatures. A fiber-optic acoustic emission measurement system was constructed to provide online monitoring of tensile damage of the woven composite laminates at cryogenic temperatures as low as 20 K. The comparations were made between the predicted cryogenic damage characteristics and in-situ AE signal analysis, assisted by scanning electron microscope (SEM) observations. The computed cryogenic damage evolution closely matched the AE signal identification results. The results indicate that fiber breakage and matrix cracking are the dominant cryogenic damage modes, and that the different low temperatures exert significant effects on the properties of the epoxy matrix, yarns and composite laminates. The combination of the AE monitoring system and the computational scheme provides a valuable tool for evaluating structural integrity and guiding the microstructural design of composite laminates used in cryogenic environments.

Development and Implementation of a New Computational Strategy for the Prediction of Elastoplastic Buckling

Development and Implementation of a New Computational Strategy for the Prediction of Elastoplastic Buckling

A Comparative Analysis of Computational Strategies in Multi-Epitope Vaccine Design Against Human Papillomavirus and Cervical Cancer.

Given the critical role of human papillomavirus (HPV) in the cause of cervical cancer and other malignancies, there is a need for innovative approaches to preventing this infection. It has been shown that immunoinformatics is an important strategy in computational vaccinology. It is used to design new multi-epitope vaccines against different types of HPV and subsequent cervical cancer. This paper reviews the scope of the entire computational pipeline of HPV vaccine design, starting from data analysis at the genomic and proteomic levels and continuing to epitope predictions of the innate and adaptive immune systems. The search strategy was based on investigating original articles published in "Google Scholar" and "PubMed" from 2015 to 2023-2024. The terms "Immunoinformatics", "Bioinformatics", "Human papillomavirus (HPV)", "Vaccine design", "In silico vaccine design", "Multi-epitope vaccine design", "Vaccinology" and "HPV vaccine" were used to for this purpose. We discussed various essential tools involved in the computational design of the vaccine process, e.g., sequence analysis, epitope prediction, conservancy analysis, tertiary structure modeling, refinement, molecular docking, molecular dynamics (MD) simulation, and in silico cloning. This review article describes immunoinformatics methods that facilitate the design of a multi-epitope vaccine against HPV. However, this pipeline can also be used to design novel chimeric vaccines for other pathogens.

Novel Combination Therapy for Heart Failure: Trimebutine-Methoxsalen Identified through Synergistic Network Virtual Screening and Experimental Validation.

Combination therapy is increasingly favored by pharmaceutical companies and researchers as an effective way to quickly discover new drugs with excellent efficacy, especially in the treatment of complex diseases. Previously, we successfully developed a computational screening method to identify such combinations, although it fell short in elucidating their synergistic mechanisms. In this work, we have transitioned to a highest single agent (HSA) synergy model for network screening, which streamlines the discovery of promising combinations and facilitates the investigation of their synergistic effects. Through this refined approach, the trimebutine-methoxsalen combination emerged as a promising candidate for heart failure (HF) treatment, exhibiting significant in vitro cardioprotective effects and effectively mitigating isoproterenol (ISO)-induced structural remodeling in the mouse heart. Further mechanistic studies have demonstrated that trimebutine and methoxsalen could synergistically inhibit intracellular calcium overload in myocardial cells and reduce the production of ROS, thus exerting cardioprotective effects. Overall, this study introduces an advanced computational strategy that not only identifies a novel combination therapy against HF but also sheds light on its underlying synergistic mechanisms.