Prosit

ProteinMPNN — deep learning-based protein sequence design from backbone structures. Uses message passing neural networks to predict amino acid sequences that fold into a given 3D backbone. Supports fi

3 shared topics • 1 shared operation

PyTorch Geometric (PyG) — graph neural network library built on PyTorch for learning on graphs and irregular structures. Provides message-passing layers (GCN, GAT, GraphSAGE, GIN, Transformer), mini-b

3 shared topics • 1 shared operation

AlphaFold2 — deep learning system for predicting 3D protein structures from amino acid sequences with atomic-level accuracy. Uses multiple sequence alignments (MSAs) and an attention-based Evoformer a

2 shared topics • 2 shared operations

Use when working with PyG (PyTorch Geometric) — the PyTorch-based library for deep learning on graphs and irregular data structures. Supports graph neural networks (GNNs) including GCN, GAT, GraphSAGE

3 shared topics

Boltz-1 — open-source deep learning model for predicting biomolecular 3D structures and interactions, approaching AlphaFold3-level accuracy. Supports protein, DNA, RNA, and small-molecule ligand struc

2 shared topics • 1 shared operation