Geometric Deep Learning on Molecular Representations.

Geometric deep learning (GDL), which is based on neural network architectures that incorporate and process symmetry information, has emerged as a recent paradigm in artificial intelligence. ... Emphasis is placed on the relevance of the learned molecular features and their complementarity to well-established molecular descriptors. ... One of the most promising advances in deep learning is geometric deep learning (GDL). ...

arXiv:2107.12375v4 fatcat:sgxlqdxiavbinly4s3zthysxbq

Open Access Multiple Versions

To facilitate the incorporation of geometry in deep learning on 3D graphs, we propose a message-passing graph convolutional network based on the distance-geometric graph representation: DG-GCN (distance-geometric ... Our work demonstrates the utility and value of DG-GCN for end-to-end deep learning on 3D graphs, particularly molecular graphs. ... Geometric Graph Representations To facilitate the incorporation of geometry in deep learning on 3D graphs, three types of geometric graph There are recent work on deep learning on 3D graphs that use the ...

arXiv:2007.03513v4 fatcat:iyhgrnbqxvd3jled3fiidv4gwy

Multiple Versions

Empirical neural scaling relations quantify the model-dependent scaling and enable optimal compute resource allocation and the identification of scalable molecular geometric deep learning model implementations ... Here, we present LitMatter, a lightweight framework for scaling molecular deep learning methods. ... Figure 1 : 1 Figure 1: Bottlenecks to scaling geometric deep learning on molecular graphs. ...

arXiv:2112.03364v1 fatcat:kgkmguoccbbihbq7rixuqbbyom

To facilitate the incorporation of geometry in deep learning on 3D graphs, we define three types of geometric graph representations: positional, angle-geometric and distance-geometric. ... Our work demonstrates the feasibility and promise of incorporating geometry, using the distance-geometric graph representation, in deep learning on 3D graphs. ... To facilitate the incorporation of geometry in deep learning on 3D graphs, we define three types of geometric graph representations: positional, angle-geometric and distance-geometric. ...

arXiv:2006.01785v1 fatcat:fedhr6cx6vgolfqewi6d7xtkwe

The incorporation of geometric properties at different levels can greatly facilitate the molecular representation learning. ... However, the fundamental property of a molecule could be altered with the augmentation method (like random perturbation) on molecular graphs. ... Geometric Learning on Molecular Graphs In the field of deep learning, geometry-based methods have shown prominent performance (Bronstein et al. 2017 ). ...

arXiv:2109.11730v1 fatcat:l2chcnrs45bexehlgvry6a2mhm

Open Access

Results We proposed Drug3D-Net, a novel deep neural network architecture based on the spatial geometric structure of molecules for predicting molecular properties. ... Compared with other deep learning methods, Drug3D-Net shows superior performance in predicting molecular properties and biochemical activities. ... Drug3D-Net is a molecular geometry-based deep neural network architecture to learn 3D drug molecular representation for property prediction tasks. ...

doi:10.1093/bib/bbab078 pmid:33822856 fatcat:l573lg5ujrcupmux2dtljsemwy

Tiered graph autoencoders provide the architecture and mechanisms for learning tiered latent representations and latent spaces for molecular graphs that explicitly represent and utilize groups (e.g., functional ... Further, tiered graph autoencoders and graph prediction together provide effective, efficient and interpretable deep learning for molecular graphs, with the former providing unsupervised, transferable ... Unsupervised Deep Learning: Tiered Graph Autoencoders The following diagram shows the tiered graph autoencoder architecture [2] for learning tiered latent representations and latent spaces: Molecular ...

arXiv:1910.11390v2 fatcat:xzllciv4qvaqfp2yjqj3dqtfqm

Multiple Versions

united under the banner of geometric deep learning. ... In this work, we present rotation- and permutation-equivariant architectures for deep learning on these small point clouds, composed of a set of products of terms from the geometric algebra and reductions ... The approach described herein applies geometric algebra to train deep learning models on point clouds, but using geometric algebra (also known as Clifford algebra) to structure the operations of neural ...

arXiv:2110.02393v1 fatcat:vlk6wbwogrer7mvh4fu72n2jde

Recent works have shown that geometric deep learning can be used on mesh-based representations of proteins to identify potential functional sites, such as binding targets for potential drugs. ... In this paper, we present a new framework for deep learning on protein structures that addresses these limitations. ... geometric deep learning. ...

doi:10.1101/2020.12.28.424589 fatcat:6fk67ycwefhk5nlkowptbdjb2y

State-of-the-art techniques based on geometric deep learning (or graph machine learning), deep generative models, reinforcement learning and knowledge graph reasoning are implemented for these tasks. ... TorchDrug benchmarks a variety of important tasks in drug discovery, including molecular property prediction, pretrained molecular representations, de novo molecular design and optimization, retrosynthsis ... The most prominent packages are PyTorch-Geometric (PyG) (Fey and Lenssen, 2019) and Deep Graph Library (DGL) (Wang et al., 2019) , which are targeted at building graph neural networks (GNNs) in PyTorch ...

arXiv:2202.08320v1 fatcat:jzbsxbhizjgwflwvh2avoyzcai

The library interfaces with popular geometric deep learning libraries: DGL, PyTorch Geometric and PyTorch3D. ... Geometric deep learning is emerging as a popular methodology in computational structural biology. ... deep learning on proteins. ...

doi:10.1101/2020.07.15.204701 fatcat:6yzfd5g7mfga5ekbb23sjozig4

As examples, we show that Molecular CT enables representational learning for molecular systems at different scales, and achieves comparable or improved results on common benchmarks using a more light-weighted ... Deep learning is changing many areas in molecular physics, and it has shown great potential to deliver new solutions to challenging molecular modeling problems. ... Representation learning on multiple levels As a GNN-based model, Molecular CT is able to perform representation learning for many-particle systems on various levels. ...

arXiv:2012.11816v2 fatcat:rtknlowbyfgb3pscyl7vw7dqpm

Open Access Multiple Versions

Here, we present a unified theoretical view of relational machine learning models which can address these tasks. ... In recent years, numerous machine learning models which attempt to solve polypharmacy side effect identification, drug-drug interaction prediction and combination therapy design tasks have been proposed ... Developing a dedicated relational machine learning framework on top of existing geometric deep learning [Fey et al., 2019; Rozemberczki et al., 2021b] and deep chemistry frameworks [Ramsundar et al. ...

arXiv:2111.02916v4 fatcat:wyzysblwqfefdemmk2dhwtlvxa

Open Access Multiple Versions

high level model training utilities, geometric deep learning, and deep chemistry layers from the PyTorch ecosystem. ... Finally, we show that ChemicalX could be used to train and score machine learning models on large drug pair datasets with hundreds of thousands of compounds on commodity hardware. ... We discuss how geometric deep learning models learn from the extracted molecular graphs, overview techniques for drug pair scoring, and open-source software for deep drug discovery. ...

arXiv:2202.05240v2 fatcat:bcgkrqharnfc3mjj4qoqpugiia

Open Access Multiple Versions

Novel emerging approaches include (i) geometric learning, i.e. learning on representations such as graphs, 3D Voronoi tessellations, and point clouds; (ii) pre-trained protein language models leveraging ... In CASP14, deep learning has boosted the field to unanticipated levels reaching near-experimental accuracy. ... The authors thank Kliment Olechnovič from Vilnius University for his help with illustrating Voronoi cells and proof-reading the manuscript, and Bowen Jing for his feedback on the manuscript. ...

arXiv:2105.07407v2 fatcat:6szubg7q2rajlj3l4vyzqri3nm

Open Access Multiple Versions

Geometric Deep Learning on Molecular Representations [article]

Preserved Fulltext

Other Versions

Distance-Geometric Graph Convolutional Network (DG-GCN) for Three-Dimensional (3D) Graphs [article]

Preserved Fulltext

Other Versions

Scalable Geometric Deep Learning on Molecular Graphs [article]

Preserved Fulltext

Geometric Graph Representations and Geometric Graph Convolutions for Deep Learning on Three-Dimensional (3D) Graphs [article]

Preserved Fulltext

GeomGCL: Geometric Graph Contrastive Learning for Molecular Property Prediction [article]

Preserved Fulltext

A spatial-temporal gated attention module for molecular property prediction based on molecular geometry

Preserved Fulltext

Deep Learning for Molecular Graphs with Tiered Graph Autoencoders and Graph Prediction [article]

Preserved Fulltext

Other Versions

Geometric Algebra Attention Networks for Small Point Clouds [article]

Preserved Fulltext

Fast end-to-end learning on protein surfaces [article]

Preserved Fulltext

TorchDrug: A Powerful and Flexible Machine Learning Platform for Drug Discovery [article]

Preserved Fulltext

Graphein - a Python Library for Geometric Deep Learning and Network Analysis on Protein Structures [article]

Preserved Fulltext

Molecular CT: Unifying Geometry and Representation Learning for Molecules at Different Scales [article]

Preserved Fulltext

Other Versions

A Unified View of Relational Deep Learning for Drug Pair Scoring [article]

Preserved Fulltext

Other Versions

ChemicalX: A Deep Learning Library for Drug Pair Scoring [article]

Preserved Fulltext

Other Versions

Protein sequence-to-structure learning: Is this the end(-to-end revolution)? [article]

Preserved Fulltext

Other Versions