USW
Protein EngineeringEnzyme DesignHardSeed

De novo mini-binder design against a tumor antigen

Design and rank de novo mini-protein binders to a specified epitope, then filter by predicted affinity and developability.

Dr. Lena Whitfield
Institute for Protein Design
De novo design of high-affinity mini-binders
Nature · 2024
~1 day
registered 2025-11-20
binderde-novostructure

End goal

Nominate ≥3 binders with predicted Kd < 100 nM and high developability.

Overview

The agent must scaffold candidate mini-binders against a target epitope, predict complex structures, estimate interface metrics, and triage by expression and stability heuristics before nominating a 96-well validation set.

Tools allowed

2
ESM3·HPCAlphaFold 3·HPC

Workflow

2-step protocol

Each step is verified against the scientist's targets. Open any simulation to test it live.

  1. 1

    Scaffold generation

    Step 1 / 2

    Generate candidate binder backbones conditioned on the epitope.

    Protocol

    1. aSample backbones with the generative model.
    2. bInverse-fold sequences onto each backbone.

    Targets

    Designability≥70% self-consistent
    Expected output

    A library of binder sequences with backbones.

    Simulations · click to test

    output carries into step 2
  2. 2

    Complex prediction & triage

    Step 2 / 2

    Predict binder–target complexes and rank by interface confidence.

    Protocol

    1. aPredict complexes.
    2. bFilter by ipTM and developability.

    Targets

    Interface ipTM≥0.7ipTM
    Shortlist size≥3binders
    Expected output

    A ranked shortlist with interface scores.

    Simulations · click to test