Review
Biochemical Research Methods
Carlos H. M. Rodrigues, Douglas E. Pires, Tom L. Blundell, David B. Ascher
Summary: Protein-protein interfaces have unique binding geometry and physicochemical properties, with concave binding sites that challenge the previous belief about their flatness. A comprehensive review of protein-protein interface landscape reveals the utilization of small binding pockets even in larger flat interfaces.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Biochemistry & Molecular Biology
Ruiyang Song, Baixin Cao, Zhenling Peng, Christopher J. Oldfield, Lukasz Kurgan, Ka-Chun Wong, Jianyi Yang
Summary: The article introduces a new sequence-based predictor, DMBS, which improves the accurate prediction of deleterious nsSNPs by optimizing conservation estimates and utilizing functional/binding residue annotations. Empirical results show that DMBS outperforms current methods in various benchmarks, demonstrating its effectiveness in guiding wet-lab experiments.
Article
Chemistry, Multidisciplinary
Ingoo Lee, Hojung Nam
Summary: Identifying drug-target interactions (DTIs) is crucial for drug discovery. Many deep learning models have been proposed to tackle the challenge of searching all drug-target spaces. However, interpretability in model construction has been neglected, which is closely related to model performance. In this study, we developed a deep learning model called HoTS, which predicts binding regions (BRs) and DTIs by training the model to predict important regions on a protein sequence. The proposed HoTS model demonstrated excellent performance in BR and DTI prediction, even without 3D structure information. The attention given to BRs and the use of transformers were found to be important for accurate prediction.
JOURNAL OF CHEMINFORMATICS
(2022)
Article
Mathematical & Computational Biology
Hanyu Luo, Wenyu Shan, Cheng Chen, Pingjian Ding, Lingyun Luo
Summary: This study presents TFBert, a variant of BERT based on pre-training, which achieved state-of-the-art results in predicting DNA-protein bindings by leveraging a task-specific pre-training strategy and large-scale multi-source DNA-protein binding data.
INTERDISCIPLINARY SCIENCES-COMPUTATIONAL LIFE SCIENCES
(2023)
Article
Biochemistry & Molecular Biology
Jiazhi Song, Guixia Liu, Jingqing Jiang, Ping Zhang, Yanchun Liang
Summary: Accurately identifying protein-ATP binding residues is crucial for protein function annotation and drug design. This paper introduces an ensemble predictor combining deep convolutional neural network and LightGBM, achieving better performance than other state-of-art prediction methods. By optimizing weight distribution, the outputs from three subclassifiers are combined to provide the final prediction result.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2021)
Article
Biochemical Research Methods
Marco Necci, Damiano Piovesan, Silvio C. E. Tosatto
Summary: Intrinsically disordered proteins present a challenge to traditional protein structure-function analysis, with computational methods, particularly deep learning techniques, showing superior performance in predicting disorder. However, predicting disordered binding regions remains difficult, and there is a significant variation in computational times among methods.
Article
Mathematical & Computational Biology
Qizhi Zhu, Lihua Wang, Ruyu Dai, Wei Zhang, Wending Tang, Yannan Bin, Zeliang Wang, Junfeng Xia
Summary: In this work, a sequence-based machine learning method named PTMC was proposed to predict the crystallization propensity of transmembrane proteins. By utilizing feature selection and extreme gradient boosting, PTMC outperformed state-of-the-art sequence-based methods on an independent test set, showing improvements in sensitivity, specificity, accuracy, MCC, and AUC compared to competitors Bcrystal and TMCrys.
INTERDISCIPLINARY SCIENCES-COMPUTATIONAL LIFE SCIENCES
(2021)
Article
Biochemistry & Molecular Biology
Chia-Tzu Ho, Yu-Wei Huang, Teng-Ruei Chen, Chia-Hua Lo, Wei-Cheng Lo
Summary: Secondary structure prediction (SSP) of proteins is an important technique in structural biology with many applications. In the past seven decades, around 300 algorithms have been published, and the estimated limit of three-state SSP accuracy has been re-evaluated to be approximately 92%, while the limit for eight-state SSP is estimated to be in the range of 84-87%. This shows that SSP remains challenging and there is room for improvement in the field.
Article
Biology
Xingyue Gu, Yijie Ding, Pengfeng Xiao
Summary: Protein sequence classification is an important field in bioinformatics that plays a vital role in functional annotation, structure prediction, and understanding protein function and interactions. However, existing machine learning methods have limitations in terms of accuracy, precision, and generalization capabilities for different types of proteins. In this study, a protein sequence classifier called MLapRVFL is proposed, which incorporates Multi-Laplacian and L2,1-norm regularization to improve the model's generalization performance, robustness, and accuracy. Experimental results demonstrate that MLapRVFL outperforms popular machine learning methods and achieves superior predictive performance compared to previous studies. Overall, the proposed MLapRVFL method makes significant contributions to protein sequence prediction.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Biotechnology & Applied Microbiology
Wei Wang, Yu Zhang, Dong Liu, HongJun Zhang, XianFang Wang, Yun Zhou
Summary: This study focuses on the identification of binding sites between DNA-binding proteins and drugs. By analyzing residue interaction network features and sequence features, a predictor for protein-drug binding sites was built. The study found that residue interaction network features can effectively describe DNA-binding proteins, and binding sites with high betweenness and high closeness values are more likely to interact with drugs.
FRONTIERS IN BIOENGINEERING AND BIOTECHNOLOGY
(2022)
Article
Biochemistry & Molecular Biology
Jianfeng Sun, Dmitrij Frishman
Summary: The study introduces a novel deep-learning approach for predicting interaction sites in transmembrane proteins, which outperforms existing methods. Results also show that approximately 10-25% of amino acid sites are predicted to be involved in interactions in the main functional families of human transmembrane proteins.
COMPUTATIONAL AND STRUCTURAL BIOTECHNOLOGY JOURNAL
(2021)
Article
Biochemical Research Methods
Zheng Jiang, Yue-Yue Shen, Rong Liu
Summary: Accurate prediction of nucleic binding residues is crucial for understanding transcription and translation processes. A novel structure-based integrative algorithm called NABind was developed, which combines a deep learning module and a template module integrated by a stacking strategy. The algorithm showed superior performance compared to traditional hybrid methods and purely deep learning-based methods.
PLOS COMPUTATIONAL BIOLOGY
(2023)
Article
Biochemical Research Methods
Yan Kang, Yulong Xu, Xinchao Wang, Bin Pu, Xuekun Yang, Yulong Rao, Jianguo Chen
Summary: Deep Learning techniques have made significant progress in protein-protein interaction (PPI) site prediction, but still have limitations. This paper introduces a novel hybrid neural network named HN-PPISP, which combines a Multi-layer Perceptron Mixer (MLP-Mixer) module and a two-stage multi-branch module for global feature extraction and PPI site prediction. Experimental results show that the model consistently achieves state-of-the-art performance over seven benchmark tests.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Biochemical Research Methods
Yan Kang, Yulong Xu, Xinchao Wang, Bin Pu, Xuekun Yang, Yulong Rao, Jianguo Chen
Summary: Motivated by the time-consuming and laborious nature of biological experimental approaches to protein-protein interaction (PPI) site prediction, a novel hybrid neural network called HN-PPISP is proposed. This model integrates a Multi-layer Perceptron Mixer (MLP-Mixer) module for local feature extraction and a two-stage multi-branch module for global feature capture. Experimental results show that HN-PPISP consistently outperforms seven baselines on real-world public datasets.
BRIEFINGS IN BIOINFORMATICS
(2023)
Article
Biochemistry & Molecular Biology
Zhonghua Wu, Sushmita Basu, Xuantai Wu, Lukasz Kurgan
Summary: This study proposes a protein sequence-based predictor for nucleic acid binding, which predicts both the binding of proteins and individual residues. The predictor offers more information than existing methods, with shorter runtime, making it particularly useful for predicting protein-NA interactions in large protein families and proteomes.
Article
Biochemistry & Molecular Biology
Attila Horvath, Monika Fuxreiter, Michele Vendruscolo, Carl Holt, John A. Carver
Summary: Casein micelles are extracellular assemblies of unstructured casein proteins stabilized by calcium phosphate nanoclusters and multivalent interactions. They can be considered as extracellular condensates formed by liquid-liquid phase separation, similar to intracellular condensates. Caseins share similarities with small heat-shock proteins, suggesting a regulatory mechanism for protein condensates.
Article
Chemistry, Multidisciplinary
Ross J. Taylor, Mauricio Aguilar Rangel, Michael B. Geeson, Pietro Sormanni, Michele Vendruscolo, Goncalo J. L. Bernardes
Summary: Post-translational protein-protein conjugation allows for the production of bioconjugates that are not achievable through genetic fusion. This study describes a method using x-clamp-mediated cysteine arylation with pentafluorophenyl sulfonamide functional groups to prepare protein-protein conjugates. By computationally designing antibodies targeting the SARS-CoV-2 receptor binding domain and using the pi-clamp sequence, dimerization was achieved, resulting in a significant increase in binding. This strategy enables the construction of molecule-protein-protein conjugates with precise chemical control over the modification sites.
JOURNAL OF THE AMERICAN CHEMICAL SOCIETY
(2022)
Article
Multidisciplinary Sciences
Zenon Toprakcioglu, Ayaka Kamada, Thomas C. T. Michaels, Mengqi Xie, Johannes Krausser, Jiapeng Wei, Andela Saric, Michele Vendruscolo, Tuomas P. J. Knowles
Summary: Primary nucleation is the fundamental event in the formation of amyloid aggregates. This study demonstrates that interfaces can modulate nucleation by affecting the primary nucleation step. The strength of surface interactions plays a crucial role in regulating nucleation rates.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2022)
Article
Multidisciplinary Sciences
Cristina Olivieri, Geoffrey C. Li, Yingjie Wang, V. S. Manu, Caitlin Walker, Jonggul Kim, Carlo Camilloni, Alfonso De Simone, Michele Vendruscolo, David A. Bernlohr, Susan S. Taylor, Gianluigi Veglia
Summary: ATP-competitive inhibitors are a major class of drugs for protein kinases. Using protein kinase A as a model system, this study shows that different ATP-competitive inhibitors modulate substrate binding cooperativity by tuning the conformational entropy of the kinase. The findings propose a new paradigm for the discovery of ATP-competitive inhibitors based on their ability to modulate the allosteric coupling between nucleotide and substrate-binding sites.
Review
Multidisciplinary Sciences
Michele Vendruscolo, Monika Fuxreiter
Summary: Condensed states of proteins play crucial roles in the organization and function of cells, and disruptions of these states can lead to various diseases. This review analyzes the identification of targets for pharmacological interventions and explores opportunities for regulating aberrant protein condensation.
NATURE COMMUNICATIONS
(2022)
Article
Biochemical Research Methods
Marc Oeller, Ryan Kang, Rosie Bell, Hannes Ausserwoger, Pietro Sormanni, Michele Vendruscolo
Summary: This article describes a computational method that incorporates the effect of pH on protein solubility predictions. The accuracy of these predictions is comparable to experimental methods, as demonstrated on various antibodies and proteins. This method, named CamSol 3.0, is now publicly available at https://www-cohsoftware.ch.cam.ac.uk/index.php/camsolph.
BRIEFINGS IN BIOINFORMATICS
(2023)
Article
Multidisciplinary Sciences
Alyssa Miller, Sean Chia, Zenon Toprakcioglu, Tuuli Hakala, Roman Schmid, Yaduo Feng, Tadas Kartanas, Ayaka Kamada, Michele Vendruscolo, Francesco Simone Ruggeri, Tuomas P. J. Knowles
Summary: In this study, a lab-on-a-chip spray approach was developed to combine rapid sample preparation, mixing, and deposition with nanoanalytical methods in chemistry and biology. This method provides enhanced spectroscopic sensitivity and single-molecule spatial resolution, allowing for multidimensional studies of heterogeneous biomolecular systems.
Review
Chemistry, Multidisciplinary
Ryan Limbocker, Nunilo Cremades, Roberta Cascella, Peter M. Tessier, Michele Vendruscolo, Fabrizio Chiti
Summary: The misfolding and aggregation of peptides and proteins into amyloid aggregates is a common feature of various protein misfolding diseases, including Alzheimer's disease and Parkinson's disease. Misfolded protein oligomers, which can form intermediates in the fibril formation process or be released by mature fibrils, are increasingly recognized as central to the development of these diseases. Despite challenges in studying these oligomers, researchers have developed methods to produce stable and reproducible populations for experimentation. These tools have provided insights into the structural determinants of oligomer toxicity and potential therapeutic strategies.
ACCOUNTS OF CHEMICAL RESEARCH
(2023)
Review
Pharmacology & Pharmacy
Michele Vendruscolo
Summary: Protein misfolding diseases, such as Alzheimer's and Parkinson's diseases, have a major impact on our healthcare systems and societies. This paper discusses drug discovery strategies to target protein misfolding and aggregation, including thermodynamic and kinetic approaches. There is a need for disease-modifying treatments to address the over 50 human disorders associated with protein misfolding and aggregation.
EXPERT OPINION ON DRUG DISCOVERY
(2023)
Review
Clinical Neurology
Mark R. Wilson, Sandeep Satapathy, Michele Vendruscolo
Summary: The proteostasis system regulates cellular processes of protein synthesis, folding, concentration, trafficking, and degradation. The mechanisms of extracellular proteostasis, particularly in the context of neurodegenerative diseases, are not well understood, but growing evidence suggests that impairment of this system may contribute to neuronal death.
NATURE REVIEWS NEUROLOGY
(2023)
Article
Biochemistry & Molecular Biology
Andras Hatos, Joao M. C. Teixeira, Susana Barrera-Vilarmau, Attila Horvath, Silvio C. E. Tosatto, Michele Vendruscolo, Monika Fuxreiter
Summary: Proteins form complex interactions in the cellular environment. The FuzPred server predicts their binding modes based on sequence without specifying the binding partners. The server also estimates the multiplicity of binding modes and visualizes different interaction behaviors on protein structures.
NUCLEIC ACIDS RESEARCH
(2023)
Article
Multidisciplinary Sciences
Alyssa Miller, Jiapeng Wei, Sarah Meehan, Christopher M. Dobson, Mark E. Welland, David Klenerman, Michele Vendruscolo, Francesco Simone Ruggeri, Tuomas P. J. Knowles
Summary: Neurodegenerative diseases such as Alzheimer's disease are caused by protein misfolding and aggregation into amyloid fibrils. This study uses atomic force microscopy and statistical theory to characterize amyloid ring structures derived from the brains of AD patients and explains the diversity in the structures formed from protein aggregation. The results show that ex vivo protofibril chains possess greater flexibility than mature amyloid fibrils, allowing them to form end-to-end connections and shedding light on their role in disease.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2023)
Article
Biochemistry & Molecular Biology
Z. Faidon Brotzakis, Thomas Lohr, Steven Truong, Samuel Hoff, Massimiliano Bonomi, Michele Vendruscolo
Summary: In recent years, cryo-electron microscopy has made significant advancements in determining biomolecular structures at the atomic level. However, studying large systems with continuous dynamics using this method has been a challenge. To address this, the metadynamic electron microscopy metainference (MEMMI) method was developed, which combines cryo-EM density maps with prior information to determine the structure and dynamics of large heterogeneous systems. This method was applied to study the amyloid fibril dynamics of the islet amyloid polypeptide (IAPP), revealing interesting characteristics of the fibril's structural variability and liquid-like dynamics in the core region.
Article
Multidisciplinary Sciences
Christine M. Lim, Alicia Gonzalez Diaz, Monika Fuxreiter, Frank W. Pun, Alex Zhavoronkov, Michele Vendruscolo
Summary: This study presents an approach that combines the PandaOmics platform with the FuzDrop method to identify disease-associated proteins prone to protein phase separation (PPS). The validated targets for Alzheimer's disease suggest the potential of this approach in identifying therapeutic targets for diseases involving dysregulation of PPS.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA
(2023)
Review
Physics, Applied
Thomas C. T. Michaels, Daoyuan Qian, Andela Saric, Michele Vendruscolo, Sara Linse, Tuomas P. J. Knowles
Summary: This Review discusses the general protein self-assembly behavior of amyloid fibrils, which is associated with functional biology and the development of diseases such as Alzheimer and Parkinson diseases. It summarizes recent progress in describing the biophysical features of amyloid self-assembly as a nucleation-and-growth process, highlighting the role of secondary nucleation. The Review also outlines non-classical aspects of aggregate formation and discusses their implications for understanding and modulating protein assembly pathways.
NATURE REVIEWS PHYSICS
(2023)