Uni-Mol: A Universal 3D Molecular Representation ...
Feb 1, 2023 · TL;DR: A universal 3D molecular pretraining framework that significantly enlarges the representation ability and application scope in drug design. Abstract: Molecular representation learning (MRL) has gained tremendous attention due to its critical role in learning from limited supervised data for applications like drug design.
Uni-Mol Block?KKN@ JKEOAO 1 7 > Fsás ? C C C O H H Atom Representation N Layers Linear Linear Linear Q K V MatMul ...
ated in Fig. 2, the Uni-Mol backbone is a Transform. r based model. It has two inputs, atom types and atom coordinates. A. d two representations (atom and pair) are maintained in the model. The atom representation is initialized from atom types, by the Embedding layer; The pair representation is initialized by inva.
NeurIPS 2023 - OpenReview
Loading. NeurIPS 2023 Conference. NeurIPS 2023 Track Datasets and Benchmarks. NeurIPS 2023 Workshop Proposals. NeurIPS 2023 Workshop. OpenReview is a long-term project to advance science through improved peer review, with legal nonprofit status through Code for Science & Society. We gratefully acknowledge the support of the OpenReview Sponsors ...
As described in UniMol [55], the encoder is a SE(3) 3D transformer that accepts tokenized atom features as input. To preserve SE(3) invariance for embedding the molecular structure, the 3D coordinate features are utilized as geometric distances. Specifically, the pairwise representationq0 ij
Uni-Mol: A Universal 3D Molecular Representation Learning …
May 16, 2022 · Herein, we propose Uni-Mol, a universal MRL framework that significantly enlarges the representation ability and application scope of MRL schemes. Uni-Mol is composed of two models with the same SE (3)-equivariant transformer architecture: a molecular pretraining model trained by 209M molecular conformations; a pocket pretraining model trained ...
Pre-training Molecular Graph Representation with 3D Geometry
Jan 28, 2022 · Molecular graph representation learning is a fundamental problem in modern drug and material discovery. Molecular graphs are typically modeled by their 2D topological structures, but it has been recently discovered that 3D geometric information plays a more vital role in predicting molecular functionalities. However, the lack of 3D information ...
OpenReview NeurIPS 2021 Summary Report
Mar 22, 2023 · In 2021 the organizers of NeurIPS (one of the flagship conferences in machine learning) decided to move from CMT to OpenReview. This report provides a summary of the NeurIPS 2021 workflow, the OpenReview services provided, the system performance, and enhancements planned for the next NeurIPS. The volume of NeurIPS paper submissions has …
3D Molecular Generation by Virtual Dynamics | OpenReview
Sep 22, 2022 · Herein, inspired by Molecular Dynamics, we propose a novel pocket-based 3D molecular generation framework VD-Gen. VD-Gen consists of a Virtual Dynamics mechanism and several carefully designed stages to generate fine-grained 3D molecules with binding positions in the pocket cavity end-to-end. Rather than directly generating or sampling atoms ...
Towards 3D Molecule-Text Interpretation in Language Models
Jan 16, 2024 · Abstract: Language Models (LMs) have greatly influenced diverse domains. However, their inherent limitation in comprehending 3D molecular structures has considerably constrained their potential in the biomolecular domain. To bridge this gap, we focus on 3D molecule-text interpretation, and propose 3D-MoLM: 3D-Molecular Language Modeling.