Built with the BoltzFold API — automated de novo binder design powered by BoltzGen and Boltz-2.

This submission demonstrates an end-to-end automated pipeline for Nipah virus binder design, fully integrated with the BoltzFold API. The pipeline generates and evaluates candidates from epitope identification through submission, showcasing the capabilities of BoltzFold's commercial-use protein folding and design platform.

Epitope generation: Penalty-driven stochastic coverage algorithm produces diverse binding sites across the target interface.

Design: BoltzGen generates de novo binder sequences for each epitope via the BoltzFold API.

Prediction & scoring: Boltz-2 predicts complex structures through the API; ipSAE is calculated locally to rank candidates.

Selection: Designs are filtered by ipSAE thresholds, competition rules (length ≤250 AA, uniqueness), and exclusion of previously submitted sequences.

Optimization: The pipeline prioritizes ipSAE (the competition metric) while considering pDockQ, ipTM, and contact quality for selection.

This approach demonstrates how the BoltzFold API enables systematic exploration of the binding landscape at scale, supporting high-throughput evaluation with automated, reproducible workflows. The pipeline is modular and checkpointed, supporting both fully automated runs and interactive analysis, all powered by BoltzFold's serverless GPU backend and multi-model routing (BoltzGen, Boltz-2, OpenFold-3).

Learn more at boltzfold.com