SHREC 2022: Protein–ligand binding site recognition

Gagliardi, Luca; Raffo, Andrea; Fugacci, Ulderico; Biasotti, Silvia; Rocchia, Walter; Hao, Huang; Amor, Boulbaba Ben; Fang, Yi; Zhang, Yuanyuan; Wang, Xiao; Christoffer, Charles; Αξενόπουλος, Απόστολος; Mylonas, Stelios K.; Daras, Petros

doi:10.1016/j.cag.2022.07.005

Cited by 13 publications

(40 citation statements)

References 44 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The binding site located on the surface of NADH oxidase is known to be hard to spot by both geometric and chemical information-driven approaches . SiteRadar (AA-specific) utilizes chemical information, optimally maps the binding site, and does not detect any false-positive pocket (Figure a).…”

Section: Resultsmentioning

confidence: 99%

“…The binding site located on the surface of NADH oxidase is known to be hard to spot by both geometric and chemical information-driven approaches. 9 SiteRadar (AAspecific) utilizes chemical information, optimally maps the binding site, and does not detect any false-positive pocket (Figure 5a). Oppositely, SiteRadar (geometric) cannot retrieve the binding site completely (Figure 5b) and only detects a part of the binding site formed by 19 grid points (the minimal size of retained pockets should be decreased by one point to retain it after clustering).…”

Section: ■ Resultsmentioning

confidence: 99%

“…Usually, the ability to detect protein− ligand pockets does not extend beyond 40% for most methods. 8 Modern approaches do not increase the prediction accuracy significantly compared with nonmachine learning algorithms, especially for the identification of previously uncharacterized, shallow solvent-exposed, 9 and allosteric binding sites 10 as well as sites of covalent inhibitors binding 11 and cavities within protein−protein interface (PPI) regions. 12 SiteRadar is a graph neural network (GNN)-based approach for identifying protein−ligand-binding sites.…”

Section: ■ Introductionmentioning

confidence: 99%

See 2 more Smart Citations

SiteRadar: Utilizing Graph Machine Learning for Precise Mapping of Protein–Ligand-Binding Sites

Evteev

Ereshchenko

Ivanenkov

2023

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Identifying ligand-binding sites on the protein surface is a crucial step in the structure-based drug design. Although multiple techniques have been proposed, including those using machine learning algorithms, the existing solutions do not provide significant advantages over nonmachine learning approaches and there is still a big room for improvement. The low ability to identify protein−ligand-binding sites makes available approaches inapplicable to automated drug design. Here, we present SiteRadar, a new algorithm for mapping cavities that are likely to bind a small-molecule ligand. SiteRadar shows higher accuracy in binding site identification compared with FPocket and PUResNet. SiteRadar demonstrates an ability to detect up to 74% of true ligand-binding sites according to the top N + 2 metric and usually covers approximately 80% of ligand atoms. Therefore, SiteRadar can be regarded as a promising solution for implementation into algorithms for automated drug design.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: ■ Resultsmentioning

confidence: 99%

Section: ■ Introductionmentioning

confidence: 99%

See 1 more Smart Citation

SiteRadar: Utilizing Graph Machine Learning for Precise Mapping of Protein–Ligand-Binding Sites

Evteev

Ereshchenko

Ivanenkov

2023

J. Chem. Inf. Model.

View full text Add to dashboard Cite

show abstract

“…Pocket detection software using an alpha sphere-based approach, such as Fpocket and Sitefinder on MOE (Chemical Computing Group Inc.), generates pseudo-atoms (alpha-spheres) based on Voronoi tessellation of the protein surface and accurately predicts druggable sites and docking spaces ( Le Guilloux et al 2009 ; Schmidtke et al 2010 ). However, it provides insufficient information on the desirable ligand size as the predicted pockets by pocket detection software are often larger than the volume of the actual bound ligands ( Supplementary Table S1 ; Gagliardi et al 2022 ). The lack of information causes a hindrance to accurate and rational drug design.…”

Section: Introductionmentioning

confidence: 99%

Pocket to concavity: a tool for the refinement of protein–ligand binding site shape from alpha spheres

et al. 2023

View full text Add to dashboard Cite

Summary Understanding the binding site of the target protein is essential for rational drug design. Pocket detection software predicts the ligand binding site of the target protein; however, the predicted protein pockets are often excessively estimated in comparison with the actual volume of the bound ligands. This study proposes a refinement tool for the pockets predicted by an alpha sphere-based approach, Pocket to Concavity (P2C). P2C is divided into two modes: Ligand-Free (LF) and Ligand-Bound (LB) modes. The LF mode provides the shape of the deep and druggable concavity where the core scaffold can bind. The LB mode searches the deep concavity around the bound ligand. Thus, P2C is useful for identifying and designing desirable compounds in Structure-Based Drug Design (SBDD). Availability and implementation Pocket to Concavity is freely available at https://github.com/genki-kudo/Pocket-to-Concavity. This tool is implemented in Python3 and Fpocket2. Supplementary information Supplementary data are available at Bioinformatics online.

show abstract

“…Backbone Models include DGCNN, Point Transformer(Zhao et al, 2021), and EGNN (Satorras et al, 2021) which have been widely used for scientific GDL(Qu & Gouskos, 2020;Atz et al, 2021;Gagliardi et al, 2022).Baseline interpretation methods include three masking-based methods BernMask, BernMask-P and PointMask, and two gradient-based methods GradGeo and GradGAM. Masking-based methods attempt to learn a mask ∈ [0, 1] for each point and may help with testing existence importance of points.…”

mentioning

confidence: 99%

Interpretable Geometric Deep Learning via Learnable Randomness Injection

Miao¹,

Luo²,

Liu³

et al. 2022

Preprint

View full text Add to dashboard Cite

Point cloud data is ubiquitous in scientific fields. Recently, geometric deep learning (GDL) has been widely applied to solve prediction tasks with such data. However, GDL models are often complicated and hardly interpretable, which poses concerns to scientists when deploying these models in scientific analysis and experiments. This work proposes a general mechanism named learnable randomness injection (LRI), which allows building inherently interpretable models based on general GDL backbones. LRI-induced models, once being trained, can detect the points in the point cloud data that carry information indicative of the prediction label. We also propose four datasets from real scientific applications that cover the domains of high energy physics and biochemistry to evaluate the LRI mechanism. Compared with previous post-hoc interpretation methods, the points detected by LRI align much better and stabler with the ground-truth patterns that have actual scientific meanings. LRI is grounded by the information bottleneck principle. LRI-induced models also show more robustness to the distribution shifts between training and test scenarios. Our code and datasets are available at https://github.com/Graph-COM/LRI.

show abstract

SHREC 2022: Protein–ligand binding site recognition

Cited by 13 publications

References 44 publications

SiteRadar: Utilizing Graph Machine Learning for Precise Mapping of Protein–Ligand-Binding Sites

SiteRadar: Utilizing Graph Machine Learning for Precise Mapping of Protein–Ligand-Binding Sites

Pocket to concavity: a tool for the refinement of protein–ligand binding site shape from alpha spheres

Interpretable Geometric Deep Learning via Learnable Randomness Injection

Contact Info

Product

Resources

About