Generating 3D molecules conditional on receptor binding sites with deep generative models

Ragoza, Matthew; Masuda, Takuya; Koes, David Ryan

doi:10.1039/d1sc05976a

Cited by 90 publications

(117 citation statements)

References 36 publications

Supporting

Mentioning

101

Contrasting

Order By: Relevance

“…The value at each grid point, or voxel, in each channel is determined by an atomic smearing function, which quantitatively defines the influence each atom has on its surrounding voxels. Generally, an atomic smearing function depends on atomic radius and assigns a larger influence value to voxels closer to any atom, exemplified by previously explored pair correlation functions and Gaussian-like densities. , The resulting value at each voxel is the summation or maximum of all influences exerted upon it by every atom in the channel (Figure b).…”

Section: Featurization Methods Of 3d Deep Generative Modelsmentioning

confidence: 99%

“…Case 3. Ragoza et al 32 have proposed a conditional VAE model combined with 3D convolutional layers to generate ligands in protein binding sites. Similar to the authors' previous approach, 31 molecules were voxelized into 3D grids with six property channels.…”

Section: Applicationsmentioning

confidence: 99%

“…The most important goal for 3D molecular generative models is the design of ligands that bind to specific targets, which is the reason for using a more challenging method instead of combining a SMILES-or graph-based model with conformation generation modules. However, most of the 3D molecular generative models reviewed above design com- 33,35 and Ragoza et al 32 fall into this category. However, their models require further structural inference, and the novel compounds generated require experimental validation.…”

Section: Applicationsmentioning

confidence: 99%

“…Generally, an atomic smearing function depends on atomic radius and assigns a larger influence value to voxels closer to any atom, exemplified by previously explored pair correlation functions 30 and Gaussian-like densities. 31,32 The resulting value at each voxel is the summation or maximum of all influences exerted upon it by every atom in the channel (Figure 1b).…”

Section: ■ Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Advances and Challenges in De Novo Drug Design Using Three-Dimensional Deep Generative Models

Xie

Wang

et al. 2022

J. Chem. Inf. Model.

View full text Add to dashboard Cite

A persistent goal for de novo drug design is to generate novel chemical compounds with desirable properties in a labor-, time-, and cost-efficient manner. Deep generative models provide alternative routes to this goal. Numerous model architectures and optimization strategies have been explored in recent years, most of which have been developed to generate two-dimensional molecular structures. Some generative models aiming at three-dimensional (3D) molecule generation have also been proposed, gaining attention for their unique advantages and potential to directly design drug-like molecules in a target-conditioning manner. This review highlights current developments in 3D molecular generative models combined with deep learning and discusses future directions for de novo drug design.

show abstract

Section: Featurization Methods Of 3d Deep Generative Modelsmentioning

confidence: 99%

Section: Applicationsmentioning

confidence: 99%

Section: Applicationsmentioning

confidence: 99%

Section: ■ Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Advances and Challenges in De Novo Drug Design Using Three-Dimensional Deep Generative Models

Xie

Wang

et al. 2022

J. Chem. Inf. Model.

View full text Add to dashboard Cite

show abstract

“…Recent years have seen significant interest in developing machine learning models to rapidly generate and screen large numbers of molecules as potential drug candidates. Different authors have employed a range of different molecular representations, including SMILES, graphs, SELFIES, and atomic density grids, and a number of different deep learning architectures, such as generative adversarial networks, variational autoencoders and recurrent neural networks . With the aim of generating molecules with an optimal set of properties, several approaches have been proposed for multiobjective optimization, including gradient descent, reinforcement learning, Bayesian optimization and particle swarm optimization …”

Section: Introductionmentioning

confidence: 99%

Incorporating Target-Specific Pharmacophoric Information into Deep Generative Models for Fragment Elaboration

Hadfield

Imrie

Merritt

et al. 2022

J. Chem. Inf. Model.

View full text Add to dashboard Cite

Despite recent interest in deep generative models for scaffold elaboration, their applicability to fragment-to-lead campaigns has so far been limited. This is primarily due to their inability to account for local protein structure or a user’s design hypothesis. We propose a novel method for fragment elaboration, STRIFE, that overcomes these issues. STRIFE takes as input fragment hotspot maps (FHMs) extracted from a protein target and processes them to provide meaningful and interpretable structural information to its generative model, which in turn is able to rapidly generate elaborations with complementary pharmacophores to the protein. In a large-scale evaluation, STRIFE outperforms existing, structure-unaware, fragment elaboration methods in proposing highly ligand-efficient elaborations. In addition to automatically extracting pharmacophoric information from a protein target’s FHM, STRIFE optionally allows the user to specify their own design hypotheses.

show abstract

Structure‐Based Drug Discovery with Deep Learning**

et al. 2023

View full text Add to dashboard Cite

Artificial intelligence (AI) in the form of deep learning has promise for drug discovery and chemical biology, for example, to predict protein structure and molecular bioactivity, plan organic synthesis, and design molecules de novo. While most of the deep learning efforts in drug discovery have focused on ligand‐based approaches, structure‐based drug discovery has the potential to tackle unsolved challenges, such as affinity prediction for unexplored protein targets, binding‐mechanism elucidation, and the rationalization of related chemical kinetic properties. Advances in deep‐learning methodologies and the availability of accurate predictions for protein tertiary structure advocate for a renaissance in structure‐based approaches for drug discovery guided by AI. This review summarizes the most prominent algorithmic concepts in structure‐based deep learning for drug discovery, and forecasts opportunities, applications, and challenges ahead.

show abstract

Generating 3D molecules conditional on receptor binding sites with deep generative models

Abstract: The goal of structure-based drug discovery is to find small molecules that bind to a given target protein. Deep learning has been used to generate drug-like molecules with certain cheminformatic...

Cited by 90 publications

References 36 publications

Advances and Challenges in De Novo Drug Design Using Three-Dimensional Deep Generative Models

Advances and Challenges in De Novo Drug Design Using Three-Dimensional Deep Generative Models

Incorporating Target-Specific Pharmacophoric Information into Deep Generative Models for Fragment Elaboration

Structure‐Based Drug Discovery with Deep Learning**

Contact Info

Product

Resources

About