Article
Pharmacology & Pharmacy
Zan Hafner Petrovski, Barbara Hribar-Lee, Zoran Bosnic
Summary: This paper proposes a novel workflow for predicting possible binding sites of a ligand on a protein surface. By combining available ligand information for similar proteins from the PDBbind and sc-PDB databases, a three-dimensional convolutional neural network is used to consider the spatial structure of a protein. The performance analysis of the model shows achieved sensitivity of 0.829, specificity of 0.98, and F1 score of 0.517, with a distance between the real and predicted centers of less than 4 Å for 54% of larger and pharmacologically relevant binding sites.
Article
Biochemical Research Methods
Kai Sun, Xiuzhen Hu, Zhenxing Feng, Hongbin Wang, Haotian Lv, Ziyang Wang, Gaimei Zhang, Shuang Xu, Xiaoxiao You
Summary: This study presented an efficient method for predicting Mg2+ and Ca2+ ligand binding sites using deep neural network algorithm, with optimized hyper-parameters and undersampling data processing method leading to superior prediction results.
BMC BIOINFORMATICS
(2022)
Article
Chemistry, Multidisciplinary
Jeevan Kandel, Hilal Tayara, Kil To Chong
Summary: This study introduces a deep learning model and a novel data cleaning process based on structural similarity for predicting protein-ligand binding sites. The method achieved better and justifiable performance when evaluating two independent datasets using metrics such as distance, volume, and proportion.
JOURNAL OF CHEMINFORMATICS
(2021)
Article
Chemistry, Medicinal
Yan Li, Zhe Zhang, Renxiao Wang
Summary: This study presents an improved version of the empirical method HydraMap, called HydraMap v.2, which utilizes statistical potentials to predict hydration sites and compute desolvation energy during protein-ligand binding. The statistical potentials for protein-water interactions were updated based on crystal protein structures, and new potentials derived from solvated structures of small organic molecules were introduced to evaluate ligand-water interactions. HydraMap v.2 successfully predicts and compares hydration sites in a binding pocket before and after ligand binding, identifying key water molecules involved in the binding process. It also demonstrates good correlation between desolvation energies and ligand binding affinities in six target proteins, providing a cost-effective solution for estimating desolvation energy and guiding lead optimization in structure-based drug discovery.
JOURNAL OF CHEMICAL INFORMATION AND MODELING
(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
Engineering, Chemical
Xinhao Che, Shiyang Chai, Zhongzhou Zhang, Lei Zhang
Summary: An improved blind docking method with a machine learning-based scoring function is proposed in this paper for the prediction of protein-ligand binding sites, and its excellent performance is demonstrated through two cases.
CHEMICAL ENGINEERING SCIENCE
(2022)
Article
Biochemical Research Methods
Zheng-Chang Lu, Fan Jiang, Yun-Dong Wu
Summary: Phosphate binding is crucial in biological processes, and accurate prediction of phosphate binding sites is challenging. The novel PBSP method, combining energy-based ligand-binding sites identification and reverse focused docking with a phosphate probe, outperforms existing predictors with high success rates and accuracy in phosphate binding modes prediction.
Article
Pharmacology & Pharmacy
Xushan Wang, Erik J. Hembre, Paul J. Goldsmith, James P. Beck, Kjell A. Svensson, Francis S. Willard, Robert F. Bruns
Summary: Researchers have identified a compound called Compound A that functions as a positive allosteric modulator (PAM) at the dopamine D1 receptor. They also synthesized a more active form of Compound A (BMS-A1) and compared it with other known D1 PAMs. The results showed that BMS-A1 enhanced the activity of other PAMs and demonstrated the presence of three non-overlapping allosteric sites on the D1 receptor.
MOLECULAR PHARMACOLOGY
(2023)
Article
Biology
Kouhei Tachibana, Aoi Fukazawa, Ryo Nishide, Takenao Ohkawa
Summary: This study explores the importance of structural similarity in protein pockets for discovering new ligands and proposes a method to search for proteins with similar pockets and extract significant spots in binding ligands. The effectiveness of the method is confirmed by classifying ligands based on the extracted significant spots in the ligand binding to the protein.
COMPUTATIONAL BIOLOGY AND CHEMISTRY
(2021)
Article
Biochemical Research Methods
Zheng Jiang, Si-Rui Xiao, Rong Liu
Summary: This study comprehensively compared the binding and nonbinding sites, as well as different categories of binding sites, in DNA and RNA. Based on structural information, a feature-based ensemble learning classifier and a template-based classifier were established and integrated into a unified prediction framework for identifying nucleic acid binding sites. Promising results were achieved, surpassing existing methods.
BRIEFINGS IN BIOINFORMATICS
(2022)
Article
Biochemistry & Molecular Biology
Sam M. Ireland, Andrew C. R. Martin
Summary: Computational models for predicting zinc binding sites are faster and more accurate, achieving high MCC, recall, and precision scores for both structure and sequence prediction. Models focusing on binding sites with four liganding residues perform particularly well. The predictors outperform other zinc binding site predictors and are accessible online.
Article
Chemistry, Multidisciplinary
Marcus Wieder, Josh Fass, John D. Chodera
Summary: The calculation of tautomer ratios of druglike molecules is crucial in computer-aided drug discovery, but accurate calculations in aqueous solution are surprisingly difficult. Current quantum chemical approaches using continuum solvent models and rigid-rotor harmonic-oscillator thermochemistry are inaccurate despite their computational expense.
Article
Biochemical Research Methods
Yajing Guo, Xiujuan Lei
Summary: In this study, we propose a novel method called circ-pSBLA for predicting the binding sites between circular RNAs and RNA binding proteins. By constructing a pseudo-Siamese framework integrating BiLSTM network and soft attention mechanism, circ-pSBLA achieves superior performance and outperforms other methods.
Article
Biochemistry & Molecular Biology
Huimin Shen, Youzhi Zhang, Chunhou Zheng, Bing Wang, Peng Chen
Summary: This paper proposes a new cascade graph-based convolutional neural network architecture for accurately predicting the binding affinity between proteins and ligands. The method deals with non-Euclidean irregular data and outperforms most current methods in experiments, showing superior performance in predicting protein-ligand binding affinity.
INTERNATIONAL JOURNAL OF MOLECULAR SCIENCES
(2021)
Article
Biochemistry & Molecular Biology
Shiwei Wang, Haoyu Lin, Zhixian Huang, Yufeng He, Xiaobing Deng, Youjun Xu, Jianfeng Pei, Luhua Lai
Summary: This study constructed the CavitySpace database, the first pocket library for all proteins in the human proteome. By analyzing known ligand binding sites, it was found that these sites can be well recovered. The predicted binding sites were grouped according to their similarity, which can be used in protein function prediction and drug repurposing studies. Novel binding sites in highly reliable predicted structure regions provide new opportunities for drug discovery.