Improving detection of protein-ligand binding sites with 3D segmentation

Stepniewska-Dziubinska, Marta M.; Zielenkiewicz, Piotr; Siedlecki, Paweł

doi:10.1038/s41598-020-61860-z

Cited by 94 publications

(119 citation statements)

References 22 publications

Supporting

Mentioning

118

Contrasting

Order By: Relevance

“…Though it was one of the best sequence-based performers, the authors note that structure-based methods still outperform sequence-based ones, even with the help of machine learning 43 . Some of the newest developments in this area include DeepSite 44 , DeepCSeqSite 45 , Kalasanty 46 , and UTProt Galaxy pipeline 47 . These methods all utilize 3D convolutional neural networks with various information involving sequence, distances, and other physicochemical parameters to characterize putative binding pockets.…”

Section: Lbs-prediction Methods Lbs-prediction Methodsmentioning

confidence: 99%

“…Kalasanty is a machine-learning method which utilizes a 3D fully convolutional neural network which characterizes protein binding pockets using physicochemical characteristics of protein atoms distributed across a 70 Å cubic grid 46 . The feature information used in their calculation describes: atom type, hybridization, number of bonds with other heavy atoms, number of bonds with other hetero atoms, encoding properties (hydrophobic, aromatic, acceptor, donor, and ring) of groups, and whether an atom belongs to a ligand or protein.…”

Section: Assessment Metrics Statistical Tests (Correlation R 2 T-tmentioning

confidence: 99%

See 1 more Smart Citation

Predicting binding sites from unbound versus bound protein structures

Clark

Orban

Carlson

2020

Sci Rep

View full text Add to dashboard Cite

We present the application of seven binding-site prediction algorithms to a meticulously curated dataset of ligand-bound and ligand-free crystal structures for 304 unique protein sequences (2528 crystal structures). We probe the influence of starting protein structures on the results of binding-site prediction, so the dataset contains a minimum of two ligand-bound and two ligand-free structures for each protein. We use this dataset in a brief survey of five geometry-based, one energy-based, and one machine-learning-based methods: Surfnet, Ghecom, LIGSITEcsc, Fpocket, Depth, AutoSite, and Kalasanty. Distributions of the F scores and Matthew’s correlation coefficients for ligand-bound versus ligand-free structure performance show no statistically significant difference in structure type versus performance for most methods. Only Fpocket showed a statistically significant but low magnitude enhancement in performance for holo structures. Lastly, we found that most methods will succeed on some crystal structures and fail on others within the same protein family, despite all structures being relatively high-quality structures with low structural variation. We expected better consistency across varying protein conformations of the same sequence. Interestingly, the success or failure of a given structure cannot be predicted by quality metrics such as resolution, Cruickshank Diffraction Precision index, or unresolved residues. Cryptic sites were also examined.

show abstract

Section: Lbs-prediction Methods Lbs-prediction Methodsmentioning

confidence: 99%

Section: Assessment Metrics Statistical Tests (Correlation R 2 T-tmentioning

confidence: 99%

Predicting binding sites from unbound versus bound protein structures

Clark

Orban

Carlson

2020

Sci Rep

View full text Add to dashboard Cite

show abstract

“…DeepSite 46 follows a similar approach to P2Rank as they use a CNN to score all points on the protein surface and clusters all points with high scores to generate candidate binding pockets. Kalasanty, 47 on the other hand, passes the entire protein structure through a CNN-based segmentation model inspired by the U-Net 48 to generate the predicted binding sites in one step. It assigns a probability to each voxel of being part of a pocket.…”

Section: Introductionmentioning

confidence: 99%

DeepPocket: Ligand Binding Site Detection and Segmentation using 3D Convolutional Neural Networks

Aggarwal¹,

Gupta²,

Chelur³

et al. 2021

Preprint

View full text Add to dashboard Cite

<div> A structure-based drug design pipeline involves the development of potential drug molecules or ligands that form stable complexes with a given receptor at its binding site. A prerequisite to this is finding druggable and functionally relevant binding sites on the 3D structure of the protein. Although several methods for detecting binding sites have been developed beforehand, a majority of them surprisingly fail in the identification and ranking of binding sites accurately. The rapid adoption and success of deep learning algorithms in various sections of structural biology beckons the usage of such algorithms for accurate binding site detection. As a combination of geometry based software and deep learning, we report a novel framework, DeepPocket that utilises 3D convolutional neural networks for the rescoring of pockets identified by Fpocket and further segments these identified cavities on the protein surface. Apart from this, we also propose another dataset SC6K containing protein structures submitted in the Protein Data Bank (PDB) from January 2018 till February 2020 for ligand binding site (LBS) detection. DeepPocket's results on various binding site datasets and SC6K highlights its better performance over current state-of-the-art methods and good generalization ability over novel structures. </div><div><br></div>

show abstract

“…Nonetheless, a variety of molecular descriptors have been identified for representing a molecule in a 3D data structure (vide infra) and successfully used for a diverse set of problems ranging from protein-ligand binding affinity prediction [2][3][4][5][6][7][8][9] and receptor binding site detection and classification [10][11][12][13] to the prediction of material properties 14,15 and NMR chemical shifts. 16 A major complication associated with 3D input representations is its high data sparsity aggravated due to its 3D grid cell structure; therefore, in this article, we advocate the use of spatially dense descriptors, especially the ones based on the electron distribution in the molecule, thus providing an alternative to mitigate the data structure sparsity.…”

Section: Introductionmentioning

confidence: 99%

3D Dense Convolutional Neural Networks Utilizing Molecular Topological Features for Accurate Atomization Energy Predictions

Gupta¹,

Raghavachari²

2021

Preprint

View full text Add to dashboard Cite

<p>Deep learning methods provide a novel way to establish a correlation between two quantities. In this context, computer vision techniques like 3D-Convolutional Neural Networks (3D-CNN) become a natural choice to associate a molecular property with its structure due to the inherent three-dimensional nature of a molecule. However, traditional 3D input data structures are intrinsically sparse in nature, which tend to induce instabilities during the learning process, which in turn may lead to under-fitted results. To address this deficiency, in this project, we propose to use quantum-chemically derived molecular topological features, namely, Localized Orbital Locator (LOL) and Electron Localization Function (ELF), as molecular descriptors, which provide a relatively denser input representation in three-dimensional space. Such topological features provide a detailed picture of the atomic configuration and inter-atomic interactions in the molecule and are thus ideal for predicting properties that are highly dependent on molecular geometry. Herein, we demonstrate the efficacy of our proposed model by applying it to the task of predicting atomization energies for the QM9-G4MP2 dataset, which contains ~134-k molecules. Furthermore, we incorporated the Δ-ML approach into our model, allowing us to reach beyond benchmark accuracy levels (~1.0 kJ mol<sup>−1</sup>).<sup> </sup>We consistently obtain impressive MAEs of the order 0.1 kcal mol<sup>−1</sup> (~ 0.42 kJ mol<sup>−1</sup>) <i>versus</i> G4(MP2) theory using relatively modest models, which could potentially be improved further using additional compute resources.</p>

show abstract

Improving detection of protein-ligand binding sites with 3D segmentation

Cited by 94 publications

References 22 publications

Predicting binding sites from unbound versus bound protein structures

Predicting binding sites from unbound versus bound protein structures

DeepPocket: Ligand Binding Site Detection and Segmentation using 3D Convolutional Neural Networks

3D Dense Convolutional Neural Networks Utilizing Molecular Topological Features for Accurate Atomization Energy Predictions

Contact Info

Product

Resources

About