2021
DOI: 10.1021/acs.jcim.0c01496
|View full text |Cite
|
Sign up to set email alerts
|

Generative Deep Learning for Targeted Compound Design

Abstract: In the past few years, de novo molecular design has increasingly been using generative models from the emergent field of Deep Learning, proposing novel compounds that are likely to possess desired properties or activities. De novo molecular design finds applications in different fields ranging from drug discovery and materials sciences to biotechnology. A panoply of deep generative models, including architectures as Recurrent Neural Networks, Autoencoders, and Generative Adversarial Networks, can be trained on… Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
4

Citation Types

0
79
0
1

Year Published

2021
2021
2024
2024

Publication Types

Select...
3
3
2
2

Relationship

0
10

Authors

Journals

citations
Cited by 107 publications
(80 citation statements)
references
References 108 publications
0
79
0
1
Order By: Relevance
“…Advances in de novo design software have been extensively reviewed, 2 and examples include both rule-based generation methods such as OpenGrowth, 3 AutoGrow, 4 and LigBuilder, 5 and recently deep generative methods for molecule design. 6 With advances like these described above, much progress has been made in the important problem of optimising a molecular design within the context of a pre-defined scoring function and binding pocket. However, whether the designed molecule indeed has high biological activity is crucially reliant on the accuracy of the methods that are used to generate and score poses of the designed molecules, as well as other assumptions, such as a rigid receptor, that might be employed.…”
Section: Introductionmentioning
confidence: 99%
“…Advances in de novo design software have been extensively reviewed, 2 and examples include both rule-based generation methods such as OpenGrowth, 3 AutoGrow, 4 and LigBuilder, 5 and recently deep generative methods for molecule design. 6 With advances like these described above, much progress has been made in the important problem of optimising a molecular design within the context of a pre-defined scoring function and binding pocket. However, whether the designed molecule indeed has high biological activity is crucially reliant on the accuracy of the methods that are used to generate and score poses of the designed molecules, as well as other assumptions, such as a rigid receptor, that might be employed.…”
Section: Introductionmentioning
confidence: 99%
“…In the scenario of designing inhibitors for a new target, a sufficient amount of exemplar molecules is required, which is likely unavailable and requires costly and time-consuming screening experiments to obtain. As the majority of existing deep generative frameworks (see Sousa, et al 5 for a review of generative deep learning for targeted molecule design) still rely on learning from targetspecific libraries of binder compounds, they limit exploration beyond a fixed library of known and monolithic molecules, while preventing generalization of the machine learning framework toward more novel targets. As a result, while some studies [6][7][8] that use deep generative models for target-specific inhibitor design have been experimentally validated, rarely have such models demonstrated sufficient versatility to be broadly deployable across dissimilar protein targets, without having access to detailed target-specific prior knowledge (e.g., target structure or binder library).…”
Section: Introductionmentioning
confidence: 99%
“…Beyond merely proposing realistic molecules, molecular design typically additionally demands the focused generation of candidates with desired properties. This can be achieved 15 by enhancing generative approaches, i.e. via transfer learning, [16][17][18][19][20] semi-supervised learning, 21,22 conditional generation, [21][22][23][24][25][26] reinforcement learning [27][28][29][30] or by carrying out optimization in a wellstructured (latent) representation space.…”
Section: Introductionmentioning
confidence: 99%