Corollary

Research

  • Papers

Library

  • Catalog

Account

  • Jobs
  • Billing
SettingsResearch demo · not for clinical or commercial use
Corollary
  1. Catalog
  2. Antibody & Biologics
  3. IgFold

Antibody & Biologics · Ruffolo et al., Johns Hopkins

IgFold

Antibody structure prediction ~1000× faster than AlphaFold2.

Academic and non-commercial use onlyThe JHU Academic License licence does not permit commercial use of this model.JHU Academic License
Availability
Not runnable yet

No backend is configured for this model.

Typical latency
2.0s
Credits per run
~14
Compute tier
A · Fast

1–30s

Not runnable yet

This model isn't available to run right now.

Run this model

IgFold

Single-letter amino acid code. Whitespace and FASTA headers are stripped automatically.

Single-letter amino acid code. Whitespace and FASTA headers are stripped automatically.

Needs heavy chain

About this model

IgFold exploits the fact that antibodies share a conserved framework: it only really has to predict the CDR loops. The result is sub-second structure prediction accurate enough for downstream docking and developability work.

Standardized I/O contract

Every model in the Hub speaks the same contract, which is what lets the Router and the agent call any of them without special-casing.

Inputs

  • heavysequence · required

    Heavy chain

  • lightsequence

    Light chain

Outputs

  • pdbstructure

    All-atom coordinates in PDB format, viewable and downloadable.

  • plddtvector

    pLDDT score per residue (0–100). Above 70 is generally reliable.

Specification

Hardware
1× A100 16GB
GPU memory
8 GB
Version
0.4
Licence
JHU Academic License
Backend
not configured
MCP server
mcp-antibody-server

Tasks

structure prediction
antibodyfoldingfast