Article
Computer Science, Artificial Intelligence
Honghong Yang, Hongxi Liu, Yumei Zhang, Xiaojun Wu
Summary: This paper proposes a hierarchical parallel multi-scale graph convolutional network (HPM-GNet) for 3D human pose estimation. The network addresses the problem of feature assimilation in graph convolutional networks and achieves state-of-the-art performance on benchmark datasets.
APPLIED SOFT COMPUTING
(2023)
Article
Computer Science, Artificial Intelligence
Wei Zhao, Hong Zhao
Summary: This paper proposes a hierarchical long-tailed classification method based on multi-granularity knowledge transfer (MGKT) to address the issue of long-tailed data classification. By utilizing a multi-granularity knowledge graph and a multi-scale feature fusion network, the proposed model outperforms other models in experiments.
PATTERN RECOGNITION
(2024)
Article
Construction & Building Technology
Yin Li, Nima Bonyadi, Ashleigh Papakyriakou, Bruno Lee
Summary: This study proposes a hierarchical decomposition approach to achieve global optimal solutions for multi-objective optimization problems in building design. By decomposing the problem into two subproblems to be solved sequentially, the method reduces computational costs and retains the full set of solutions. In a case study, the approach successfully decreased the number of simulations while maintaining the original problem's full set of solutions.
JOURNAL OF BUILDING ENGINEERING
(2021)
Article
Remote Sensing
Tang Liu, Ling Yao, Jun Qin, Ning Lu, Hou Jiang, Fan Zhang, Chenghu Zhou
Summary: In this study, a Multi-Scale Geoscience Network (MS-GeoNet) is proposed for automatic building footprint extraction. The proposed method optimizes for the multi-scale nested characteristics and spatial correlation between buildings and surroundings in remote sensing images, resulting in significant improvements in accuracy compared to traditional methods.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Multidisciplinary Sciences
Zhikun Cheng, Liejun Wang
Summary: Medical image segmentation provides effective methods for accuracy and robustness of organ segmentation, lesion detection, and classification. Fusing rich multi-scale features can augment segmentation accuracy. Global and local information are critical for segmentation results, considering the density of diseased tissue and the importance of multi-scale, global, and local information.
SCIENTIFIC REPORTS
(2023)
Article
Thermodynamics
L. Bartolucci, S. Cordiner, V. Mulone, S. Pasquale, A. Sbarra
Summary: Multi energy systems are effective in moving towards a decentralized low-carbon system, with hydrogen being a promising energy carrier. Research shows that building-scale configurations can achieve simultaneous environmental and economic benefits, and the integration of hydrogen technologies and storage systems is an effective solution for reducing energy consumption and carbon dioxide emissions.
Article
Biochemical Research Methods
Xue Li, Peifu Han, Wenqi Chen, Changnan Gao, Shuang Wang, Tao Song, Muyuan Niu, Alfonso Rodriguez-Paton
Summary: This study proposes a protein-protein interaction (PPI) prediction model called multi-scale architecture residual network for PPIs (MARPPI) that utilizes dual-channel and multi-feature methods. The model leverages Res2vec to obtain residue association information and uses various descriptors to capture amino acid composition and order information, physicochemical properties, and information entropy. MARPPI achieves high accuracy rates ranging from 91.80% to 99.01% on different datasets, outperforming other advanced methods. The results also demonstrate the model's ability to detect hidden interactions and predict cross-species interactions.
BRIEFINGS IN BIOINFORMATICS
(2023)
Article
Computer Science, Artificial Intelligence
Cheng Zhao, Bei Xia, Weiling Chen, Libao Guo, Jie Du, Tianfu Wang, Baiying Lei
Summary: The study proposed a pediatric echocardiographic segmentation method combining MS-Net, BFF-Net, and W-Unet, which significantly improved the diagnostic capability for congenital heart diseases.
APPLIED SOFT COMPUTING
(2021)
Article
Computer Science, Information Systems
Yan Zhang, Xudong Zhou, Ke Wang, Nian Wang, Zenghui Li
Summary: We propose a multi-scale hierarchical node graph neural network (MsHN-GNN) that enhances the representation ability of node features by constructing hierarchical nodes with multi-scale information through deep extraction of structural features, global features, and local features of samples. MsHN-GNN sorts the hierarchical relationship among sub-nodes and updates the hierarchical nodes. Experimental results demonstrate that the model outperforms other few-shot learning methods in classification performance.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Yitong Yang, Yongjun Zhang, Zhongwei Cui, Zhi Li, Yujie Xu, Haoliang Zhao, Yangtin Ou, Heliang Yang, Xihe Wang
Summary: The proposed method, MMAFN, tackles the limitations of rain streak removal task by introducing a two-stage network structure and incorporating joint channel coordinate attention block, inception attention branch block, and multi-level feature fusion block. Extensive experiments demonstrate the superiority of the network, with an average PSNR improvement of 0.2dB across five test sets and up to 0.93dB improvement on the Test100 test set.
INTERNATIONAL JOURNAL OF INTELLIGENT SYSTEMS
(2022)
Article
Computer Science, Artificial Intelligence
Yitang Wang, Kunpeng Li, Qingye Li, Yong Pang, Liye Lv, Wei Sun, Xueguan Song
Summary: Multi-fidelity information fusion has attracted increasing attention for its promising in engineering design and optimization. In order to improve the approximation performance, a feature mapping-based hierarchical surrogate model for the fusion of LF and HF information is developed. The experiments demonstrate that our method outperforms other methods in 97.436% of the thirteen test problems under three different quantitative evaluations.
KNOWLEDGE-BASED SYSTEMS
(2023)
Article
Materials Science, Composites
Tuo Zhang, Yuping Duan, Jiangyong Liu, Huifang Pang, Lingxi Huang, Xinran Ma, Yupeng Shi, Ben Ma
Summary: This study reports hierarchical metamaterials constructed from pyramid metastructure composites and Fibonacci spiral elements, with a wide effective absorption bandwidth. The use of multi-scale optimization and metastructure design is effective in improving multiple resonance thickness and wavefront transmission. The hybrid absorber exhibits robustness with incident angle across transverse magnetic polarization.
COMPOSITES SCIENCE AND TECHNOLOGY
(2022)
Article
Environmental Sciences
Lihua Jian, Shaowu Wu, Lihui Chen, Gemine Vivone, Rakiba Rayhana, Di Zhang
Summary: This paper proposes a multi-scale and multi-stream fusion network (MMFN) that leverages the multi-scale information of the source images to improve the spatial resolution of a multi-spectral image while preserving spectral signatures. The proposed approach outperforms state-of-the-art methods according to qualitative and quantitative assessment on three real remote sensing datasets (i.e., QuickBird, Pleiades, and WorldView-2).
Article
Environmental Sciences
Geding Yan, Haitao Jing, Hui Li, Huanchao Guo, Shi He
Summary: Deep learning algorithms provide an effective solution to the inefficiencies and poor results of traditional methods for building footprint extraction from high-resolution remote sensing imagery. To address the challenges in building extraction, we propose the multi-scale boundary-refined HRNet (MBR-HRNet) model, which preserves detailed boundary features for accurate building segmentation.
Article
Computer Science, Artificial Intelligence
Ruining Tang, Zhenyu Liu, Yiguo Song, Guifang Duan, Jianrong Tan
Summary: An increasing number of researchers are using deep-learning-based object detection methods for visual defect detection in industrial manufacturing. However, large-scale variation in defect detection hinders further improvement in accuracy. Therefore, a hierarchical multi-scale block (HMS-Block) is proposed to provide scale-abundant features for multi-scale defect detection. The proposed HMSNet shows high practical utility in industrial applications and outperforms existing methods in terms of detection accuracy and efficiency.
JOURNAL OF INTELLIGENT MANUFACTURING
(2023)
Article
Biochemistry & Molecular Biology
Maximilian M. Sauer, M. Alejandra Tortorici, Young-Jun Park, Alexandra C. Walls, Leah Homad, Oliver J. Acton, John E. Bowen, Chunyan Wang, Xiaoli Xiong, Willem de van der Schueren, Joel Quispe, Benjamin G. Hoffstrom, Berend-Jan Bosch, Andrew T. McGuire, David Veesler
Summary: The study identified a monoclonal antibody, B6, that can cross-react with eight beta-coronavirus spike proteins from three viral lineages, targeting a conserved cryptic epitope for potential broad coronavirus neutralization, marking a potential breakthrough in pan-beta-coronavirus vaccine design.
NATURE STRUCTURAL & MOLECULAR BIOLOGY
(2021)
Article
Chemistry, Multidisciplinary
Moumita Ray, Giorgia Brancolini, David C. Luther, Ziwen Jiang, Roberto Cao-Milan, Alejandro M. Cuadros, Andrew Burden, Vincent Clark, Subinoy Rana, Rubul Mout, Ryan F. Landis, Stefano Corni, Vincent M. Rotello
Summary: Control over supramolecular recognition between proteins and nanoparticles is crucial for therapeutic applications and sensor development. Most current binding approaches use tags, but this study presents a new strategy of co-engineering nanoparticles and proteins for high affinity binding. This strategy maintains high binding affinity even under physiological conditions.
Article
Multidisciplinary Sciences
Young-Jun Park, Anna De Marco, Tyler N. Starr, Zhuoming Liu, Dora Pinto, Alexandra C. Walls, Fabrizia Zatta, Samantha K. Zepeda, John Bowen, Kaitlin S. Sprouse, Anshu Joshi, Martina Giurdanella, Barbara Guarino, Julia Noack, Rana Abdelnabi, Shi-Yan Caroline Foo, Florian A. Lempp, Fabio Benigni, Gyorgy Snell, Johan Neyts, Sean Pj Whelan, Herbert W. Virgin, Jesse D. Bloom, Davide Corti, Matteo Samuele Pizzuto, David Veesler
Summary: Understanding broadly neutralizing sarbecovirus antibody responses is crucial for developing countermeasures against SARS-CoV-2 variants and future sarbecoviruses. The human monoclonal antibody S2K146 has been found to broadly neutralize ACE2-binding viruses and shows potential for clinical development. Conserved ACE2-binding residues could be leveraged for developing vaccines that elicit broad immunity.
Article
Multidisciplinary Sciences
Danny D. Sahtoe, Florian Praetorius, Alexis Courbet, Yang Hsia, Basile I. M. Wicky, Natasha Edman, Lauren M. Miller, Bart J. R. Timmerman, Justin Decarreau, Hana M. Morris, Alex Kan, Asim K. Bera, David Baker
Summary: This study demonstrates the successful generation of stable, folded, and soluble designs using implicit negative design, which can undergo subunit exchange to enable reconfiguration of various complexes. The approach provides a general route to designing asymmetric reconfigurable protein systems.
Article
Biochemistry & Molecular Biology
M. Alejandra Tortorici, Alexandra C. Walls, Anshu Joshi, Young-Jun Park, Rachel T. Eguia, Marcos C. Miranda, Elizabeth Kepl, Annie Dosey, Terry Stevens-Ayers, Michael J. Boeckh, Amalio Telenti, Antonio Lanzavecchia, Neil P. King, Davide Corti, Jesse D. Bloom, David Veesler
Summary: The isolation of CCoV-HuPn-2018 from a child respiratory swab indicates the presence of a previously underappreciated number of coronaviruses infecting humans. This study determined the structures of the virus's spike glycoprotein and its binding to various receptors, as well as the cross-neutralizing activity of antibodies. These findings provide a foundation for the development of vaccines and treatments against this zoonotic pathogen.
Article
Surgery
Natasha I. Edman, Sara L. Zettervall, Maianna N. Dematteis, Amir Ghaffarian, Sherene Shalhub, Matthew P. Sweet
Summary: This study assessed the gender-related differences in clinical comorbidities, anatomic suitability, and frailty among an unselected cohort of patients with TAAA5. The study found that women with TAAA5 had higher metrics of frailty and anatomic risk that were not captured by comorbidity-based risk assessments, which could explain the lower intervention rates for women with TAAA5.
JOURNAL OF VASCULAR SURGERY
(2022)
Article
Multidisciplinary Sciences
Jedediah M. Singer, Scott Novotney, Devin Strickland, Hugh K. Haddox, Nicholas Leiby, Gabriel J. Rocklin, Cameron M. Chow, Anindya Roy, Asim K. Bera, Francis C. Motta, Longxing Cao, Eva-Maria Strauch, Tamuka M. Chidyausiku, Alex Ford, Ethan Ho, Alexander Zaitzeff, Craig O. Mackenzie, Hamed Eramian, Frank DiMaio, Gevorg Grigoryan, Matthew Vaughn, Lance J. Stewart, David Baker, Eric Klavins
Summary: In this study, a high-throughput assay was used to evaluate the stability of a large number of novel proteins, and neural network models were built to predict protein stability and generate new protein sequences. The results showed that, despite limitations, the predictive model can significantly improve protein stability.
Article
Multidisciplinary Sciences
A. Courbet, J. Hansen, Y. Hsia, N. Bethel, Y-J Park, C. Xu, A. Moyer, S. E. Boyken, G. Ueda, U. Nattermann, D. Nagarajan, D-A Silva, W. Sheffler, J. Quispe, A. Nord, N. King, P. Bradley, D. Veesler, J. Kollman, D. Baker
Summary: This study explores the construction of protein machinery by designing axle and rotor components, achieving successful assembly in vitro and in vivo, and demonstrating a new approach to protein structure construction.
Article
Cell Biology
Andrew C. Hunt, James Brett Case, Young-Jun Park, Longxing Cao, Kejia Wu, Alexandra C. Walls, Zhuoming Liu, John E. Bowen, Hsien-Wei Yeh, Shally Saini, Louisa Helms, Yan Ting Zhao, Tien-Ying Hsiang, Tyler N. Starr, Inna Goreshnik, Lisa Kozodoy, Lauren Carter, Rashmi Ravichandran, Lydia B. Green, Wadim L. Matochko, Christy A. Thomson, Bastian Vogeli, Antje Kruger, Laura A. VanBlargan, Rita E. Chen, Baoling Ying, Adam L. Bailey, Natasha M. Kafai, Scott E. Boyken, Ajasja Ljubetic, Natasha Edman, George Ueda, Cameron M. Chow, Max Johnson, Amin Addetia, Mary-Jane Navarro, Nuttada Panpradist, Michael Gale, Benjamin S. Freedman, Jesse D. Bloom, Hannele Ruohola-Baker, Sean P. J. Whelan, Lance Stewart, Michael S. Diamond, David Veesler, Michael C. Jewett, David Baker
Summary: Researchers used a cell-free expression workflow to screen and optimize constructs containing computationally designed miniprotein inhibitors of SARS-CoV-2. They found that a homotrimeric version of the ACE2 mimic AHB2 (TRI2-2) achieved the broadest efficacy. The TRI2-2 miniprotein neutralized Omicron, Delta, and other variants with greater potency than clinically used monoclonal antibodies. It also provided prophylactic and therapeutic protection against SARS-CoV-2 in mice when administered intranasally. The designed miniprotein receptor mimics offer a widely applicable antiviral therapeutic strategy with advantages over antibodies and native receptor traps.
SCIENCE TRANSLATIONAL MEDICINE
(2022)
Article
Multidisciplinary Sciences
Young-Jun Park, Dora Pinto, Alexandra C. Walls, Zhuoming Liu, Anna De Marco, Fabio Benigni, Fabrizia Zatta, Chiara Silacci-Fregni, Jessica Bassi, Kaitlin R. Sprouse, Amin Addetia, John E. Bowen, Cameron Stewart, Martina Giurdanella, Christian Saliba, Barbara Guarino, Michael A. Schmid, Nicholas M. Franko, Jennifer K. Logue, Ha V. Dang, Kevin Hauser, Julia di Iulio, William Rivera, Gretja Schnell, Anushka Rajesh, Jiayi Zhou, Nisar Farhat, Hannah Kaiser, Martin Montiel-Ruiz, Julia Noack, Florian A. Lempp, Javier Janer, Rana Abdelnabi, Piet Maes, Paolo Ferrari, Alessandro Ceschi, Olivier Giannini, Guilherme Dias De Melo, Lauriane Kergoat, Herve Bourhy, Johan Neyts, Leah Soriaga, Lisa A. Purcell, Gyorgy Snell, Sean P. J. Whelan, Antonio Lanzavecchia, Herbert W. Virgin, Luca Piccoli, Helen Y. Chu, Matteo Samuele Pizzuto, Davide Corti, David Veesler
Summary: SARS-CoV-2 Omicron sublineages have spike mutations that allow them to evade antibodies from previous infection or vaccination. Hybrid immunity or booster shots can generate neutralizing antibodies against Omicron variants, and breakthrough infections lead to the production of neutralizing antibodies in the nasal mucosa. Antibodies derived from memory B cells or plasma cells of Omicron breakthrough cases show cross-reactivity with different receptor-binding domains, while primary Omicron infections elicit B cells with narrow specificity. A highly potent pan-variant-neutralizing antibody has been identified as a potential candidate for clinical development.
Article
Biochemistry & Molecular Biology
Sherry Bermeo, Andrew Favor, Ya-Ting Chang, Andrew Norris, Scott E. Boyken, Yang Hsia, Hugh K. Haddox, Chunfu Xu, T. J. Brunette, Vicki H. Wysocki, Gira Bhabha, Damian C. Ekiert, David Baker
Summary: The de novo design of protein chains that associate to form a heterotrimer is a challenging task in protein design. In this study, we designed helical heterotrimers with buried hydrogen bond networks and large aromatic residues. We obtained ten designs in which all three chains assembled into heterotrimers. Crystal structures and electron microscopy confirmed the successful assembly of these designed structures, which could be used for constructing larger protein nanostructures and modulating cell signaling.
NATURE STRUCTURAL & MOLECULAR BIOLOGY
(2022)
Article
Multidisciplinary Sciences
Amin Addetia, Luca Piccoli, James Brett Case, Young-Jun Park, Martina Beltramello, Barbara Guarino, Ha Dang, Guilherme Dias de Melo, Dora Pinto, Kaitlin Sprouse, Suzanne M. Scheaffer, Jessica Bassi, Chiara Silacci-Fregni, Francesco Muoio, Marco Dini, Lucia Vincenzetti, Rima Acosta, Daisy Johnson, Sambhavi Subramanian, Christian Saliba, Martina Giurdanella, Gloria Lombardo, Giada Leoni, Katja Culap, Carley Mcalister, Anushka Rajesh, Exequiel Dellota, Jiayi Zhou, Nisar Farhat, Dana Bohan, Julia Noack, Alex Chen, Florian A. Lempp, Joel Quispe, Lauriane Kergoat, Florence Larrous, Elisabetta Cameroni, Bradley Whitener, Olivier Giannini, Pietro Cippa, Alessandro Ceschi, Paolo Ferrari, Alessandra Franzetti-Pellanda, Maira Biggiogero, Christian Garzoni, Stephanie Zappi, Luca Bernasconi, Min Jeong Kim, Laura E. Rosen, Gretja Schnell, Nadine Czudnochowski, Fabio Benigni, Nicholas Franko, Jennifer K. Logue, Courtney Yoshiyama, Cameron Stewart, Helen Chu, Herve Bourhy, Michael A. Schmid, Lisa A. Purcell, Gyorgy Snell, Antonio Lanzavecchia, Michael S. Diamond, Davide Corti, David Veesler
Summary: The recently emerged BQ.1.1 and XBB.1.5 variants of SARS-CoV-2 have a higher affinity for the host ACE2 receptor and more efficiently promote fusion with host cell membranes compared to earlier Omicron variants. Although the neutralizing activity is reduced, vaccine-induced human plasma antibodies still cross-react with and trigger effector functions against current Omicron variants, providing a mechanism of protection against disease.