Initiative BySimBioSys

GLACIER is a community-facing platform that provides computational tools and infrastructure for the analysis and modeling of glycans, glycoproteins, and other glycoconjugates. It integrates physics-based simulations, data-driven methods, and structural analyses for scalable research across glycoscience and immunology.

VASCO Analysis

Viral Antibody Structural Complex Analysis - AI-powered interface prediction

1
2

Upload Structures for VASCO

Upload antibody and antigen PDB files. VASCO uses MSA-powered neural networks to predict interface residues.

Required files:

  • Antibody PDB: Must contain Light (L) and Heavy (H) chains
  • Antigen PDB: One or more antigen chains
  • Both files must be in standard PDB format

Processing Time & Results

VASCO analysis typically completes within 4-8 hours. This includes MSA generation, deep learning inference, and result visualization. A results link will be provided immediately upon submission where you can check the status and access your results once ready.

Upload PDB file containing antibody light and heavy chains

Drop antibody PDB here or click to browse

Supports .pdb files only

Upload PDB file containing antigen chain(s)

Drop antigen PDB here or click to browse

Supports .pdb files only

Upload both antibody and antigen PDB files to continue

1

Upload Files

Provide antibody and antigen PDB structures with chain IDs

2

Deep Learning

MSA generation + graph neural networks predict interface residues

3

Get Results

Receive predictions and visualizations via email in 4-8 hours

Citation

Cite VASCO

If you use VASCO in your research, please cite our work to support continued development.

@software{vasco2025,
  title = {{VASCO: Viral Antibody Structural Complex Analysis via GLACIER Platform}},
  author = {{SimBioSys Lab}},
  organization = {Northeastern University},
  year = {2025},
  url = {https://glacier-simbiosys.com/vasco},
  note = {MSA-powered antibody interface prediction}
}

About VASCO

VASCO (Viral Antibody Structural Complex Analysis) uses state-of-the-art deep learning:

  • Multiple Sequence Alignments (MSA) for evolutionary information via HHblits
  • ESM2 protein language model embeddings for sequence context
  • Graph Neural Networks for spatial structure representation
  • Transformer attention for sequence-structure integration

Processing time: 4-8 hours | GPU-accelerated inference | Per-residue confidence scores

For questions about citations or to report issues, please contact simbiosyslab.neu@gmail.com