GNINA 1.0: molecular docking with deep learning

McNutt, Andrew; Francoeur, Paul; Aggarwal, Rishal; Masuda, Takuya; Meli, Rocco; Ragoza, Matthew; Sunseri, Jocelyn

doi:10.1186/s13321-021-00522-2

Cited by 310 publications

(406 citation statements)

References 52 publications

Supporting

Mentioning

342

Contrasting

Order By: Relevance

“…Additional file 1 : Figure S5 compares per-class Pearson’s correlation coefficient obtained with AEScore (and reported in Fig. 4 ) using results obtained with GNINA [ 20 , 21 ], a CNN-based scoring function. We see that for most classes the Pearson’s correlation coefficient obtained with both methods is similar.…”

Section: Resultsmentioning

confidence: 99%

“…Additional file 1 : Figure S11 shows the variation in CNN-based predictions as a function of the angle of rotation for a particular complex. Data augmentation with random translations and rotations has proved to be essential to prevent overfitting and significantly improve training in CNN-based scoring functions [ 20 , 21 ], but this is computationally expensive—another advantage of our approach.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Learning protein-ligand binding affinity with atomic environment vectors

et al. 2021

Self Cite

View full text Add to dashboard Cite

Scoring functions for the prediction of protein-ligand binding affinity have seen renewed interest in recent years when novel machine learning and deep learning methods started to consistently outperform classical scoring functions. Here we explore the use of atomic environment vectors (AEVs) and feed-forward neural networks, the building blocks of several neural network potentials, for the prediction of protein-ligand binding affinity. The AEV-based scoring function, which we term AEScore, is shown to perform as well or better than other state-of-the-art scoring functions on binding affinity prediction, with an RMSE of 1.22 pK units and a Pearson’s correlation coefficient of 0.83 for the CASF-2016 benchmark. However, AEScore does not perform as well in docking and virtual screening tasks, for which it has not been explicitly trained. Therefore, we show that the model can be combined with the classical scoring function AutoDock Vina in the context of $$\Delta$$ Δ -learning, where corrections to the AutoDock Vina scoring function are learned instead of the protein-ligand binding affinity itself. Combined with AutoDock Vina, $$\Delta$$ Δ -AEScore has an RMSE of 1.32 pK units and a Pearson’s correlation coefficient of 0.80 on the CASF-2016 benchmark, while retaining the docking and screening power of the underlying classical scoring function.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Discussionmentioning

confidence: 99%

Learning protein-ligand binding affinity with atomic environment vectors

et al. 2021

Self Cite

View full text Add to dashboard Cite

show abstract

“…Accurate identification of near-native binding poses from decoy poses is a prerequisite for many downstream simulation tasks, such as binding affinity prediction and SBVS. Over the last few years, a number of MLSFs for binding pose prediction have been reported [ 26 , 36 – 46 ]. Some of them were trained to explicitly predict the root-mean-square-deviation (RMSD) values of binding poses, while the others were trained to directly distinguish near-native poses from high-RMSD ones.…”

Section: Introductionmentioning

confidence: 99%

“…Very recently, Francoeur et al reported a standardized dataset named CrossDocked2020 set with 22.5 million poses generated by docking ligands into multiple similar binding pockets to better mimic the real-world scenarios, and they comprehensively estimated the scoring and docking powers of their grid-based CNN models [ 42 ]. Based on the dataset and assessment results, they further released the 1.0 version of GNINA, which could be considered as the first publicly available docking software that integrated an ensemble of CNNs as a SF [ 46 ]. However, it seems that the dataset may be not so suitable for the large-scale assessment of SFs due to its complexity and randomness.…”

Section: Introductionmentioning

confidence: 99%

The impact of cross-docked poses on performance of machine learning classifier for protein–ligand binding pose prediction

Shen

Gao

et al. 2021

J Cheminform

View full text Add to dashboard Cite

Structure-based drug design depends on the detailed knowledge of the three-dimensional (3D) structures of protein–ligand binding complexes, but accurate prediction of ligand-binding poses is still a major challenge for molecular docking due to deficiency of scoring functions (SFs) and ignorance of protein flexibility upon ligand binding. In this study, based on a cross-docking dataset dedicatedly constructed from the PDBbind database, we developed several XGBoost-trained classifiers to discriminate the near-native binding poses from decoys, and systematically assessed their performance with/without the involvement of the cross-docked poses in the training/test sets. The calculation results illustrate that using Extended Connectivity Interaction Features (ECIF), Vina energy terms and docking pose ranks as the features can achieve the best performance, according to the validation through the random splitting or refined-core splitting and the testing on the re-docked or cross-docked poses. Besides, it is found that, despite the significant decrease of the performance for the threefold clustered cross-validation, the inclusion of the Vina energy terms can effectively ensure the lower limit of the performance of the models and thus improve their generalization capability. Furthermore, our calculation results also highlight the importance of the incorporation of the cross-docked poses into the training of the SFs with wide application domain and high robustness for binding pose prediction. The source code and the newly-developed cross-docking datasets can be freely available at https://github.com/sc8668/ml_pose_prediction and https://zenodo.org/record/5525936, respectively, under an open-source license. We believe that our study may provide valuable guidance for the development and assessment of new machine learning-based SFs (MLSFs) for the predictions of protein–ligand binding poses.

show abstract

“…In addition to virtual screening, attempts have been made to generate molecules by directly optimizing a docking score [ 19 , 20 , 21 , 22 ]. Recently, Boitreaud et al [ 23 ] suggested the OptiMol approach based on binding energy optimization for drug design using a generative model and docking using adaptive sampling (CbAS) [ 24 ] to maximize the objective function.…”

Section: Introductionmentioning

confidence: 99%

V-Dock: Fast Generation of Novel Drug-like Molecules Using Machine-Learning-Based Docking Score and Molecular Optimization

Choi

Lee

2021

IJMS

View full text Add to dashboard Cite

We propose a computational workflow to design novel drug-like molecules by combining the global optimization of molecular properties and protein-ligand docking with machine learning. However, most existing methods depend heavily on experimental data, and many targets do not have sufficient data to train reliable activity prediction models. To overcome this limitation, protein-ligand docking calculations must be performed using the limited data available. Such docking calculations during molecular generation require considerable computational time, preventing extensive exploration of the chemical space. To address this problem, we trained a machine-learning-based model that predicted the docking energy using SMILES to accelerate the molecular generation process. Docking scores could be accurately predicted using only a SMILES string. We combined this docking score prediction model with the global molecular property optimization approach, MolFinder, to find novel molecules exhibiting the desired properties with high values of predicted docking scores. We named this design approach V-dock. Using V-dock, we efficiently generated many novel molecules with high docking scores for a target protein, a similarity to the reference molecule, and desirable drug-like and bespoke properties, such as QED. The predicted docking scores of the generated molecules were verified by correlating them with the actual docking scores.

show abstract

GNINA 1.0: molecular docking with deep learning

Cited by 310 publications

References 52 publications

Learning protein-ligand binding affinity with atomic environment vectors

Learning protein-ligand binding affinity with atomic environment vectors

The impact of cross-docked poses on performance of machine learning classifier for protein–ligand binding pose prediction

V-Dock: Fast Generation of Novel Drug-like Molecules Using Machine-Learning-Based Docking Score and Molecular Optimization

Contact Info

Product

Resources

About