相关参考文献
注意:仅列出部分参考文献,下载原文获取全部文献信息。
Editorial Material
Physiology
Andrew D. McCulloch
Summary: Combining machine learning and mechanistic computational modeling allows for the discovery of new genotype-phenotype relationships in heart disease.
JOURNAL OF GENERAL PHYSIOLOGY
(2023)
Article
Materials Science, Multidisciplinary
Andrew J. Lew et al.
Summary: Inspired by natural materials, this paper presents a deep learning approach using an attention-based diffusion model to predict and design hierarchical architected materials with tailored mechanical properties. Experiments on a four-hierarchy level materials system demonstrate that the model offers efficient forward and inverse performance across the entire range of deformation and rapidly discovers multiple solutions that satisfy design objectives.
Article
Materials Science, Multidisciplinary
Markus J. Buehler
Summary: In this study, a computational approach for analyzing and designing multiscale architected materials is presented. The challenge lies in effectively modeling complex multi-level material structures for hierarchical design approaches. The authors propose an integrated deep neural network architecture that learns coarse-grained representations of complex microstructure data and solves forward and inverse problems through an attention-based diffusion model. The application of the method in the analysis and design of highly porous metamaterials is demonstrated, and the mechanical behavior is validated using molecular dynamics simulations.
MODELLING AND SIMULATION IN MATERIALS SCIENCE AND ENGINEERING
(2023)
Article
Chemistry, Multidisciplinary
Bo Ni et al.
Summary: We present two generative deep-learning models capable of predicting amino acid sequences and 3D protein structures based on secondary structure design objectives. Both models are robust to imperfect inputs and have the capacity for de novo design, enabling the discovery of novel protein sequences not found in natural mechanisms or systems. The residue-level secondary structure design model demonstrates higher accuracy and more diverse sequences. These findings highlight untapped opportunities for protein design beyond known proteins. Our models, trained on a dataset extracted from experimentally known 3D protein structures using an attention-based diffusion model, have potential applications in the generative design of various biological or engineering systems. Further research could explore additional conditioning and other functional properties of the generated proteins beyond structural objectives.
Review
Chemistry, Multidisciplinary
Ady Suwardi et al.
Summary: Biomaterials research has historically been hindered by long development periods, but the application of machine learning in materials science has greatly accelerated progress. The combination of machine learning with high-throughput theoretical predictions and experiments has shifted the traditional trial and error paradigm to a data-driven paradigm, which is driving the discovery and application of biomaterials.
ADVANCED MATERIALS
(2022)
Article
Materials Science, Biomaterials
Chi-Hua Yu et al.
Summary: This article reports a deep learning model that predicts the secondary structure content of proteins from their primary sequences. The model, using convolutional and recurrent architectures, accurately predicts the content of alpha-helix and beta-sheet structures. The predictions show excellent agreement with newly identified protein structures and have the potential for rapidly designing proteins with specific secondary structures.
ACS BIOMATERIALS SCIENCE & ENGINEERING
(2022)
Article
Nanoscience & Nanotechnology
Yu-Chuan Hsu et al.
Summary: This paper describes a method for generating 3D architected materials based on mathematically parameterized human readable word input. The researchers used a combination of generative adversarial networks and language-image pre-training neural networks to generate images, which were then translated into 3D architectures and 3D printed. This method has broad applications and can provide new avenues for the analysis and manufacturing of architected materials.
Article
Materials Science, Multidisciplinary
Markus J. Buehler
Summary: Attention-based transformer neural networks have a significant impact in modeling physical systems. The FieldPerceiver model, using a multi-headed self-attention approach, is capable of accurately predicting physical field data and material properties, demonstrating its broad generalization capacity. The model's advantages include scalability, no domain knowledge requirement, and the ability to capture crack problems in fracture mechanics.
Article
Nanoscience & Nanotechnology
Xi Chen et al.
Summary: In this paper, the numerical performance of classical conforming finite element schemes for the time-dependent incompressible Navier-Stokes equations is improved by adding dissipation, inspired by physics. The dissipative terms, constructed through the discontinuity of numerical quantities across interior edges, decouple the space and time discretizations and are motivated by energy stability and error equations associated with the unsteady problem. Additionally, the added dissipation can be viewed as an alternative for the grad-div stabilization from a physical approach, within the framework of the variational multiscale, and shows competitive performance in reducing numerical errors compared to other conventional stabilizations.
Article
Mechanics
Markus J. Buehler
Summary: Dynamic fracture is a significant area of materials analysis, and a machine learning model derived from atomistic simulations can effectively describe the dynamics and key aspects of fracture. The model, trained on a small dataset, offers a rapid assessment of dynamic fracture mechanisms for complex geometries and performs well on various validation cases.
JOURNAL OF APPLIED MECHANICS-TRANSACTIONS OF THE ASME
(2022)
Article
Chemistry, Multidisciplinary
Yiwen Hu et al.
Summary: This study presents a series of models that enable high-throughput nanodynamical property predictions of proteins directly from the amino acid sequence. By utilizing neural networks and graph-based methods, these models provide atomistically based mechanistic predictions of key protein mechanical features. The graph-based transformer model performs the best among the four models, but requires a graph structure as input. On the other hand, the LSTM and transformer models offer end-to-end sequence to-property prediction capabilities, offering efficient avenues for protein engineering, analysis, and design.
Article
Biochemistry & Molecular Biology
Mohammad Madani et al.
Summary: Protein solubility is crucial for both protein function and production yield, but current experimental methods are costly and have low success rates. To improve accuracy in predicting protein solubility, a novel deep learning sequence-based predictor, DSResSol, has been developed. This predictor outperforms existing models, identifying key amino acids and dipeptides contributing to solubility prediction.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2021)
Article
Biochemical Research Methods
Zongyang Du et al.
Summary: The trRosetta server is a web-based platform for fast and accurate protein structure prediction using deep learning and Rosetta, which distinguishes itself from other similar servers in terms of rapid and accurate prediction of novel structures.
Article
Chemistry, Multidisciplinary
Jianwen Chen et al.
Summary: This study developed a new structure-aware method GraphSol for predicting protein solubility using attentive graph convolutional network, constructing a protein topology attribute graph from the sequence. The model showed superior performance and stability, being the first to utilize GCN for sequence-based protein solubility predictions.
JOURNAL OF CHEMINFORMATICS
(2021)
Article
Materials Science, Multidisciplinary
Zhenze Yang et al.
Summary: This passage explains how Transformer neural networks are used in the design of materials, combining text input and image generation to achieve advanced material design. Through iterative processes with language-image pre-training and VQGAN neural networks, images related to text prompts are successfully generated and transformed into implementable 3D models, resulting in actual text-based material samples.
FRONTIERS IN MATERIALS
(2021)
Article
Engineering, Biomedical
Kun Xue et al.
Summary: Biomaterials research can accelerate success through machine learning techniques, leading to faster commercialization. Transitioning from unstructured empirical approaches to data-driven development strategies can bring about many potential benefits.
MATERIALS TODAY BIO
(2021)
Article
Engineering, Mechanical
Kai Guo et al.
EXTREME MECHANICS LETTERS
(2020)
Article
Chemistry, Multidisciplinary
Philippe Schwaller et al.
ACS CENTRAL SCIENCE
(2019)
Review
Biochemistry & Molecular Biology
Diego Lopez Barreiro et al.
MACROMOLECULAR BIOSCIENCE
(2019)
Review
Engineering, Biomedical
Yasser Aboelkassem et al.
CURRENT OPINION IN BIOMEDICAL ENGINEERING
(2019)
Review
Health Care Sciences & Services
Mark Alber et al.
NPJ DIGITAL MEDICINE
(2019)
Article
Chemistry, Medicinal
Daniel Merk et al.
MOLECULAR INFORMATICS
(2018)
Review
Polymer Science
Shengjie Ling et al.
PROGRESS IN POLYMER SCIENCE
(2018)
Review
Nanoscience & Nanotechnology
Shengjie Ling et al.
NATURE REVIEWS MATERIALS
(2018)
Article
Multidisciplinary Sciences
Mariya Popova et al.
Article
Biochemical Research Methods
Sheng Wang et al.
PLOS COMPUTATIONAL BIOLOGY
(2017)
Article
Biology
F. Martinez-Martinez et al.
COMPUTERS IN BIOLOGY AND MEDICINE
(2017)
Article
Theodor Ackbarow et al.
Journal of Computational and Theoretical Nanoscience
(2016)
Article
Materials Science, Multidisciplinary
Tristan Giesa et al.
ADVANCED ENGINEERING MATERIALS
(2012)
Review
Engineering, Biomedical
Greta Gronau et al.
Article
Biochemistry & Molecular Biology
Robbie P. Joosten et al.
NUCLEIC ACIDS RESEARCH
(2011)
Article
Multidisciplinary Sciences
David I. Spivak et al.
Article
Materials Science, Multidisciplinary
Markus J. Buehler et al.