4,004 Hits in 7.6 sec

Pre-training Molecular Graph Representation with 3D Geometry [article]

Shengchao Liu, Hanchen Wang, Weiyang Liu, Joan Lasenby, Hongyu Guo, Jian Tang
2022 arXiv   pre-print
GraphMVP effectively learns a 2D molecular graph encoder that is enhanced by richer and more discriminative 3D geometry.  ...  To cope with this challenge, we propose the Graph Multi-View Pre-training (GraphMVP) framework where self-supervised learning (SSL) is performed by leveraging the correspondence and consistency between  ...  We note that privileged information is only used in training, while it is not available in testing. This perfectly matches our intuition of pre-training molecular representation with 3D geometry.  ... 
arXiv:2110.07728v2 fatcat:3wcbxyrzrzdg7eahhcg4zdqt5e

ChemRL-GEM: Geometry Enhanced Molecular Representation Learning for Property Prediction [article]

Xiaomin Fang, Lihang Liu, Jieqiong Lei, Donglong He, Shanzhuo Zhang, Jingbo Zhou, Fan Wang, Hua Wu, Haifeng Wang
2021 arXiv   pre-print
However, existing GNNs and pre-training strategies usually treat molecules as topological graph data without fully utilizing the molecular geometry information.  ...  To this end, we propose a novel Geometry Enhanced Molecular representation learning method (GEM) for Chemical Representation Learning (ChemRL).  ...  • We evaluated ChemRL-GEM thoroughly on various molecular property prediction datasets.  ... 
arXiv:2106.06130v3 fatcat:kmvhdzcmhfd5rdmggelphl6s6m

3D Infomax improves GNNs for Molecular Property Prediction [article]

Hannes Stärk, Dominique Beaini, Gabriele Corso, Prudencio Tossou, Christian Dallago, Stephan Günnemann, Pietro Liò
2022 arXiv   pre-print
We propose pre-training a model to reason about the geometry of molecules given only their 2D molecular graphs.  ...  During fine-tuning on molecules with unknown geometry, the GNN still generates implicit 3D information and can use it to improve downstream tasks.  ...  3D Infomax Acknowledgments The authors express their gratitude to Simon Axelrod (GEOM Dataset and molecular physics advice), Limei Wang and Meng Liu (Spherical Message Passing), Adrien Bardes and Samuel  ... 
arXiv:2110.04126v4 fatcat:dyawphieivb5vj6ujxpbis2hz4

3D Graph Contrastive Learning for Molecular Property Prediction [article]

Kisung Moon, Sunyoung Kwon
2022 arXiv   pre-print
We propose a novel contrastive learning framework, small-scale 3D Graph Contrastive Learning (3DGCL) for molecular property prediction, to solve the above problems. 3DGCL learns the molecular representation  ...  SSL where the computing resource is insufficient. (2) In most cases, they do not utilize 3D structural information for molecular representation learning.  ...  To verify the effectiveness of geometric information on molecular representation learning, we compare 3DGCL with no pre-trained model.  ... 
arXiv:2208.06360v2 fatcat:6fpv3ei23zbaxltraultuwhoza

Molecule3D: A Benchmark for Predicting 3D Geometries from Molecular Graphs [article]

Zhao Xu, Youzhi Luo, Xuan Zhang, Xinyi Xu, Yaochen Xie, Meng Liu, Kaleb Dickerson, Cheng Deng, Maho Nakata, Shuiwang Ji
2021 arXiv   pre-print
Here, we propose to predict the ground-state 3D geometries from molecular graphs using machine learning methods.  ...  Experimental results show that, compared with generating 3D geometries with RDKit, our method can achieve comparable prediction accuracy but with much smaller computational costs.  ...  Then, this pre-trained model with knowledge of 3D geometries can help predict antiviral activity against SARS-CoV-2.  ... 
arXiv:2110.01717v1 fatcat:jsh353cvdvd5nevzm4ncqoumwm

Triangular Contrastive Learning on Molecular Graphs [article]

MinGyu Choi, Wonseok Shin, Yijingxiu Lu, Sun Kim
2022 arXiv   pre-print
Recent contrastive learning methods have shown to be effective in various tasks, learning generalizable representations invariant to data augmentation thereby leading to state of the art performances.  ...  Regarding the multifaceted nature of large unlabeled data used in self-supervised learning while majority of real-word downstream tasks use single format of data, a multimodal framework that can train  ...  A.1 Pre-training on Large Dataset To assess the effects of unlabeled dataset size for pre-training, we pre-train TriCL with larger numbers of molecules in GEOM.  ... 
arXiv:2205.13279v1 fatcat:iaxyxqj2yfa33ko4st6gxr5dpi

Graph Neural Networks for Molecules [article]

Yuyang Wang, Zijie Li, Amir Barati Farimani
2022 arXiv   pre-print
Graph neural networks (GNNs), which are capable of learning representations from graphical data, are naturally suitable for modeling molecular systems.  ...  Many researches design GNN architectures to effectively learn topological information of 2D molecule graphs as well as geometric information of 3D molecular systems.  ...  All three methods conduct contrastive pre-training between 2D topological graphs and 3D geometric structures to learn molecular representations with 3D information embedded.  ... 
arXiv:2209.05582v1 fatcat:zldtilzujvbtzb45monkbb6fxq

Do Large Scale Molecular Language Representations Capture Important Structural Information? [article]

Jerret Ross, Brian Belgodere, Vijil Chenthamarakshan, Inkit Padhi, Youssef Mroueh, Payel Das
2021 arXiv   pre-print
Experiments show that the learned molecular representation outperforms supervised and unsupervised graph neural net baselines on several regression and classification tasks from 10 benchmark datasets,  ...  Recently, pre-trained transformer-based language models on large unlabeled corpus have produced state-of-the-art results in many downstream natural language processing tasks.  ...  Our experiments on QM9 and other benchmark datasets show, for the first time, that pre-trained transformer encoders of molecular SMILES perform competitively with geometry and graph-based NNs, suggesting  ... 
arXiv:2106.09553v2 fatcat:q4ombefdfndgrpv2io5vmun5ua

Graph-based Molecular Representation Learning [article]

Zhichun Guo, Bozhao Nan, Yijun Tian, Olaf Wiest, Chuxu Zhang, Nitesh V. Chawla
2022 arXiv   pre-print
In this survey, we systematically review these graph-based molecular representation techniques. Specifically, we first introduce the data and features of the 2D and 3D graph molecular datasets.  ...  Molecular representation learning (MRL) is a key step to build the connection between machine learning and chemical science.  ...  Datasets and Benchmarks We summarize representative molecular representation learning algorithms in pre-training, and contrastive learning.  ... 
arXiv:2207.04869v1 fatcat:eweobqwm55ahdnj2n2umdfdexa

Molecular Geometry Pretraining with SE(3)-Invariant Denoising Distance Matching [article]

Shengchao Liu, Hongyu Guo, Jian Tang
2022 arXiv   pre-print
on 2D molecular graphs.  ...  Pretraining molecular representations is critical in a variety of applications in drug and material discovery due to the limited number of labeled molecules, yet most of existing work focuses on pretraining  ...  manner [26, 27, 33] . 3D Molecular Graph (Geometry) Self-Supervised Learning.  ... 
arXiv:2206.13602v1 fatcat:xxvhqv3tezcfzhgn6eut24iqxi

Pre-training via Denoising for Molecular Property Prediction [article]

Sheheryar Zaidi, Michael Schaarschmidt, James Martens, Hyunjik Kim, Yee Whye Teh, Alvaro Sanchez-Gonzalez, Peter Battaglia, Razvan Pascanu, Jonathan Godwin
2022 arXiv   pre-print
In this paper, we describe a pre-training technique that utilizes large datasets of 3D molecular structures at equilibrium to learn meaningful representations for downstream tasks.  ...  Many important problems involving molecular property prediction from 3D structures have limited data, posing a generalization challenge for neural networks.  ...  Unlike existing pre-training methods, which largely focus on 2D graphs, our approach targets the setting where the downstream task involves 3D point clouds defining the molecular structure.  ... 
arXiv:2206.00133v1 fatcat:nywadsmrmbfklec5rpkknfpovy

3DLinker: An E(3) Equivariant Variational Autoencoder for Molecular Linker Design [article]

Yinan Huang, Xingang Peng, Jianzhu Ma, Muhan Zhang
2022 arXiv   pre-print
We compare our model with multiple conditional generative models modified from other molecular design tasks and find that our model has a significantly higher rate in recovering molecular graphs, and more  ...  graph variational autoencoder.  ...  In the 3D linker design, two fragments are defined by two unlinked subgraphs G F = (G F,1 , G F,2 ) with geometry R F , and a linker is denoted by G L with geometry R L .  ... 
arXiv:2205.07309v1 fatcat:ofrkitukljf3tng5q4nlhcaxy4

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

Elodie Laine, Stephan Eismann, Arne Elofsson, Sergei Grudinin
2021 arXiv   pre-print
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  ...  attention; (iii) equivariant architectures preserving the symmetry of 3D space; (iv) use of large meta-genome databases; (v) combinations of protein representations; (vi) and finally truly end-to-end  ...  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

ComENet: Towards Complete and Efficient Message Passing for 3D Molecular Graphs [article]

Limei Wang, Yi Liu, Yuchao Lin, Haoran Liu, Shuiwang Ji
2022 arXiv   pre-print
Many real-world data can be modeled as 3D graphs, but learning representations that incorporates 3D information completely and efficiently is challenging.  ...  Our method guarantees full completeness of 3D information on 3D graphs by achieving global and local completeness. Notably, we propose the important rotation angles to fulfill global completeness.  ...  By doing this, the complete representation for a whole 3D molecular graph is eventually achieved.  ... 
arXiv:2206.08515v3 fatcat:whjtme2eqjel5gh3lvk3rylmku

An End-to-End Framework for Molecular Conformation Generation via Bilevel Programming [article]

Minkai Xu, Wujie Wang, Shitong Luo, Chence Shi, Yoshua Bengio, Rafael Gomez-Bombarelli, Jian Tang
2021 arXiv   pre-print
Predicting molecular conformations (or 3D structures) from molecular graphs is a fundamental problem in many applications.  ...  Specifically, the molecular graph is first encoded in a latent space, and then the 3D structures are generated by solving a principled bilevel optimization program.  ...  For the input features of the graph representation, we only derive the atom and bond types from molecular graphs.  ... 
arXiv:2105.07246v2 fatcat:auxvaup7nvbljeold4kqczw7om
« Previous Showing results 1 — 15 out of 4,004 results