Article
Biochemistry & Molecular Biology
Maxat Kulmanov, Fernando Zhapa-Camacho, Robert Hoehndorf
Summary: DeepGOWeb is a protein function prediction method based on deep learning and sequence similarity, which provides accurate and fast predictions through a website, API, and SPARQL query language. It ensures predicted functions are consistent with the Gene Ontology and can provide predictions for any protein and any function.
NUCLEIC ACIDS RESEARCH
(2021)
Article
Biochemistry & Molecular Biology
Hao Tian, Sian Xiao, Xi Jiang, Peng Tao
Summary: Allostery is a biological process in which an effector modulator binds to a protein at a site distant from the active site. Identifying allosteric sites is crucial for allosteric drug development. We developed PASSer, a web application that predicts and visualizes allosteric sites, and it has been widely used in over 70 countries.
NUCLEIC ACIDS RESEARCH
(2023)
Article
Computer Science, Theory & Methods
A-Long Jin, Wenchao Xu, Song Guo, Bing Hu, Kwan Yeung
Summary: This article introduces a new Parameter Server framework that allows for overlapping communication and computation in distributed training, reducing waiting time. The framework also offers a solution to the straggler problem, minimizing the impact of stale updates on global parameters.
IEEE TRANSACTIONS ON PARALLEL AND DISTRIBUTED SYSTEMS
(2022)
Article
Biochemistry & Molecular Biology
Laiyi Fu, Yingxin Cao, Jie Wu, Qinke Peng, Qing Nie, Xiaohui Xie
Summary: UFold is a deep learning-based method for RNA secondary structure prediction, which uses a novel image-like representation of RNA sequences to achieve accurate predictions in a short time, outperforming previous methods on within-family datasets and showing similar performance on distinct RNA families.
NUCLEIC ACIDS RESEARCH
(2022)
Article
Computer Science, Interdisciplinary Applications
Xiaodong Yang, Xiaoping Wu, Mingxin Yue
Summary: In this study, an algorithm for 3-D TEM modeling for loop-source devices was developed and verified. The research found that the modeling accuracy is more sensitive to space meshing, and the proposed algorithm is packed as open-source software TEMF3DT.
COMPUTERS & GEOSCIENCES
(2022)
Article
Biochemical Research Methods
Amelie Barozet, Kevin Molloy, Marc Vaisset, Christophe Zanon, Pierre Fauret, Thierry Simeon, Juan Cortes
Summary: MoMA-LoopSampler is a sampling method that explores the conformational space of flexible protein loops globally, using a large structural library and reinforcement-learning-based approach. It generates a set of statistically likely loop states satisfying geometric constraints and can sample experimentally observed conformations.
Article
Computer Science, Interdisciplinary Applications
Gregoire Uhlrich, Farvah Mahmoudi, Alexandre Arbey
Summary: In the future, studies beyond the standard model will be increasingly important due to the growing amount of data, but manual predictions in these models are impractical. MARTY is a new C++ framework that automates calculations and simplifies physical quantities, aiding in BSM research.
COMPUTER PHYSICS COMMUNICATIONS
(2021)
Article
Computer Science, Interdisciplinary Applications
Feng Feng, Yi-Fan Xie, Qiu-Chen Zhou, Shan-Rong Tang
Summary: HepLib is a C++ library for computations in High Energy Physics which combines well-known packages like qgraf, FORM, and FIRE or KIRA for high efficiency. Its core feature lies in numerical evaluation of master integrals using sector decomposition, providing a new implementation in C++ with many features.
COMPUTER PHYSICS COMMUNICATIONS
(2021)
Article
Computer Science, Interdisciplinary Applications
Feng Feng, Shan-Rong Tang, Ya-Di Gao
Summary: HepLib is a C++ library for high energy physics computations, it utilizes other established libraries including GiNaC, Fermat, Form, Fire, etc. The latest version introduces the use of Flint for simplification of multivariate polynomials, providing higher performance through thread-based parallelism for polynomial evaluations.
COMPUTER PHYSICS COMMUNICATIONS
(2023)
Article
Biochemistry & Molecular Biology
Brian Jimenez-Garcia, Jorge Roel-Touris, Didier Barradas-Bautista
Summary: The LightDock Server is a web server for integrative modeling of macromolecular interactions, offering ease of use and improved user experience.
NUCLEIC ACIDS RESEARCH
(2023)
Article
Biochemical Research Methods
Daniel T. Rademaker, Kevin J. van Geemen, Li C. Xue
Summary: Computational simulations such as molecular dynamics and docking play a crucial role in understanding protein dynamics and interaction conformations, along with experimental methods. GradPose is introduced as a fast and memory-efficient tool for superimposing structures generated by large-scale simulations. It outperforms traditional methods in terms of speed and memory requirement reduction, especially for larger protein structures. The pre-determined residue-residue correspondence is a prerequisite for GradPose.
Article
Engineering, Mechanical
Wei-Wei Huang, Linlin Li, Zhiwei Zhu, Li-Min Zhu
Summary: This paper proposes a comprehensive dynamics model based on the principles of magnetic equivalent circuits for the design and control of the normal-stressed electromagnetic actuated FTS. The feasibility of the model is verified through finite element analysis. The effectiveness of the proposed model and methods is experimentally validated.
MECHANICAL SYSTEMS AND SIGNAL PROCESSING
(2022)
Article
Operations Research & Management Science
Sigrun Andradottir, Hayriye Ayhan, Douglas G. Down
Summary: This article explores scenarios where synchronous resource assignment is necessary or desirable, and introduces a queueing network model with flexible servers to address this issue. By introducing the concept of configuration, the effects of resource synchronization can be determined in a unified way. The maximal capacity of the system is determined by solving a linear programming problem, and this information is used to construct policies with high capacity. The article compares synchronous server assignment with asynchronous approaches and highlights the advantages of the synchronous approach in terms of applicability, implementation, and capacity. Several examples are provided to illustrate the modeling framework.
Article
Computer Science, Artificial Intelligence
Jiemin Fang, Yuzhu Sun, Qian Zhang, Kangjian Peng, Yuan Li, Wenyu Liu, Xinggang Wang
Summary: This paper introduces a Fast Network Adaptation (FNA++) method that adjusts pre-trained networks for segmentation and detection tasks, leading to improved performance. The method outperforms existing networks in semantic segmentation, object detection, and human pose estimation, both manually and by NAS. Furthermore, FNA++ significantly reduces computation cost compared to state-of-the-art segmentation and detection NAS approaches.
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE
(2021)
Article
Computer Science, Information Systems
Yong Chen, Yuquan Zhu, Haifeng Chen, Yan Shen, Zhao Xu
Summary: Studies have shown that the newly proposed NewLearn++.NSE-Error-based algorithm can further improve the accuracy of the ensemble classifier under the same time complexity as the Learn++.NSE algorithm, making it suitable for fast classification learning of long-term accumulated big data.
Article
Biochemical Research Methods
Jose Ramon Lopez-Blanco, Pablo Chacon
Article
Biochemical Research Methods
Maria Kadukova, Karina Dos Santos Machado, Pablo Chacon, Sergei Grudinin
Summary: This study introduces a novel coarse-grained potential based on a 3D joint probability distribution function that depends only on the pairwise orientation and position between protein backbone and ligand atoms. Despite its simplicity, this approach demonstrates competitive results in docking and screening tasks compared to state-of-the-art scoring functions.
Article
Chemistry, Multidisciplinary
Roberto Melero, Carlos Oscar S. Sorzano, Brent Foster, Jose-Luis Vilas, Marta Martinez, Roberto Marabini, Erney Ramirez-Aportela, Ruben Sanchez-Garcia, David Herreros, Laura del Cano, Patricia Losana, Yunior C. Fonseca-Reyna, Pablo Conesa, Daniel Wrapp, Pablo Chacon, Jason S. McLellan, Hemant D. Tagare, Jose-Maria Carazo
Article
Biochemical Research Methods
Chloe Quignot, Pierre Granger, Pablo Chacon, Raphael Guerois, Jessica Andreani
Summary: This study focuses on integrating atomic-level evolutionary information into scoring functions to improve protein docking accuracy. By incorporating evolutionary information from a small number of homologous sequences, success rates are significantly increased. The best homology-enriched score reaches a success rate of 34.4%, while a consensus approach combining multiple homology-enriched scores further boosts the success rate to 40%.
Article
Biochemistry & Molecular Biology
Chloe Quignot, Guillaume Postic, Helene Bret, Julien Rey, Pierre Granger, Samuel Murail, Pablo Chacon, Jessica Andreani, Pierre Tuffery, Raphael Guerois
Summary: The InterEvDock3 protein docking server utilizes evolutionary constraints to generate structural models of protein assemblies, providing 10 candidate complexes and interface predictions. Three key innovations were implemented to improve model reliability, with server performance validated on large benchmark databases.
NUCLEIC ACIDS RESEARCH
(2021)
Article
Multidisciplinary Sciences
Maria Teresa Bueno-Carrasco, Jorge Cuellar, Marte I. Flydal, Cesar Santiago, Trond-Andre Krakenes, Rune Kleppe, Jose R. Lopez-Blanco, Miguel Marcilla, Knut Teigen, Sara Alvira, Pablo Chacon, Aurora Martinez, Jose M. Valpuesta
Summary: Tyrosine hydroxylase (TH) is a key enzyme in the synthesis of dopamine (DA) and other catecholamines. This study used Cryo-EM to determine the structures of human TH with and without DA, as well as S40 phosphorylated TH. The study revealed the inhibitory and stabilizing effects of DA, as well as the counteraction by S40 phosphorylation, providing insights into the regulatory mechanisms of TH and dopamine homeostasis.
NATURE COMMUNICATIONS
(2022)
Article
Chemistry, Medicinal
Jose Ramon Lopez-Blanco, Yves Dehouck, Ugo Bastolla, Pablo Chacon
Summary: We propose a novel method based on constrained normal mode analysis in internal coordinates to efficiently explore local protein loop conformations. By reducing the manifold of possible loop configurations to a set of orthogonal motions encoding concerted rotations of all backbone dihedral angles, our approach can effectively explore the conformational space of closed loops. Validation results demonstrate its sampling power on protein loops with highly variable experimental structures, and show acceptable resemblance with ensembles generated by long molecular simulations on exposed and diverse loops. Compared to other methods, the lack of restrictions on the number of dihedrals that can be altered simultaneously is a major advantage, along with the computational efficiency achieved by requiring diagonalization of a small matrix and energetic contextualization of motion modes using the elastic network model.
JOURNAL OF CHEMICAL INFORMATION AND MODELING
(2022)
Article
Mathematics, Applied
Alberto Pepe, Joan Lasenby, Pablo Chacon
Summary: This article discusses solving rotation problems in computer vision through deep learning, comparing the use of bivector rotation parameterization and 6D continuous representation to improve regression accuracy and reconstruction precision. Research shows that using bivector rotation parameterization outperforms the 6D representation, with better robustness and fewer learning parameters.
MATHEMATICAL METHODS IN THE APPLIED SCIENCES
(2022)
Article
Biochemical Research Methods
Ivan Martin Hernandez, Yves Dehouck, Ugo Bastolla, Jose Ramon Lopez-Blanco, Pablo Chacon
Summary: This article introduces a structure-based stability prediction method upon mutation, which is important for protein engineering and design, as well as understanding genetic diseases or drug resistance events. The method uses a simple residue-based orientational potential model and only requires parameterizing 12 amino acid-dependent weights for stability prediction. The method, called KORPM, accurately predicts mutational effects on an independent benchmark dataset, whether the wild-type or mutated structure is used as starting point. Compared with state-of-the-art methods, KORPM achieves better results in terms of root mean square error, correlation, receiver operating characteristics, and precision-recall curves.
Article
Biochemistry & Molecular Biology
Amelie Barozet, Pablo Chacon, Juan Cortes
Summary: Loops in protein structures are crucial for various biological functions, but their conformational variability poses a challenge for structural investigation. This paper provides an overview of current computational approaches to flexible loop modeling, emphasizing the difficulty in accurately capturing the conformational variability of long flexible loops. Future advancements in this field are expected to come from a combination of experimental and computational techniques, leading to a better understanding of the relationships between loop sequence, structural flexibility, and functional roles. Accurate loop modeling holds promise for applications in biomedicine and biotechnology.
CURRENT RESEARCH IN STRUCTURAL BIOLOGY
(2021)
Meeting Abstract
Biophysics
I. Martin Hernandez, J. R. Lopez-Blanco, P. Chacon
EUROPEAN BIOPHYSICS JOURNAL WITH BIOPHYSICS LETTERS
(2019)
Meeting Abstract
Biophysics
J. R. Lopez-Blanco, P. Chacon
EUROPEAN BIOPHYSICS JOURNAL WITH BIOPHYSICS LETTERS
(2019)