Loading Video...
NTHRYS
Arrow

Graph Neural Networks for Structural Biology Training | 3D Equivariant GNNs for Pockets, Interfaces & Affinity

NTHRYS >> Services >> Academic Services >> Training Programs >> Bioinformatics Training >> Proteomics, Structural Bioinformatics, Molecular Modeling >> Graph Neural Networks for Structural Biology Training | 3D Equivariant GNNs for Pockets, Interfaces & Affinity

Graph Neural Networks for Structural Biology — Hands-on

Learn geometric deep learning for proteins, ligands, and complexes. Represent structures as graphs and surfaces, train 2D/3D equivariant GNNs for pocket/interaction/affinity tasks, and ship reproducible evaluation reports with clear baselines and error bars.

Graph Neural Networks for Structural Biology
Help Desk · WhatsApp
Session 1
Fee: Rs 6300
Molecular/Protein Graphs & GNN Basics
  • Graphs for chemistry & proteins: atoms/residues, bonds/contacts, kNN & radius graphs
  • Theory
  • Message passing & attention; node/edge/global features
  • PyTorch Geometric DGL GraphConv/GAT/GIN
  • Featurization & loaders: PDB/mmCIF → graph datasets
  • AA props/SS/ASA ligand/metal flags
Session 2
Fee: Rs 8400
Protein Geometry: 3D, Surfaces & Equivariance
  • 3D coordinates, frames & invariances; surface meshes & point clouds
  • triangulated surfaces MaSIF-like features
  • E (n) /SE (3) -equivariant GNNs for 3D proteins/complexes
  • EGNN/SE (3) -Transformer SchNet/DimeNet TFN (overview)
  • Efficiency & stability: batching, cutoffs, neighbor lists
  • mixed precision checkpointing
Session 3
Fee: Rs 11200
Tasks: Pockets, Interfaces, Rescoring & Affinity
  • Pocket/surface site prediction & residue-level function
  • classification/segmentation imbalanced data
  • PPI & protein–ligand: interface prediction, docking rescoring, affinity regression
  • PDBbind SAbDab/PP interfaces PR-AUC/RMSE/CI
  • Generalization & controls: scaffold/time splits, leakage checks
  • bootstraps calibration
Session 4
Fee: Rs 14000
Mini Capstone: Train → Evaluate → Report
  • Build a small GNN (PyG/DGL) for a chosen task (e.g., docking pose rescoring or pocket detection)
  • Theory + Practical
  • Benchmarking & deployment: ONNX/TorchScript export, inference profiling
  • baselines & CIs error analysis
  • Deliverables: figures, tables, README & manifest
  • PDF CSV/TSV checksums


PDF