pxdesign

What problem does it solve?

De novo protein binder design is computationally intensive and requires integrated modeling, filtering, and validation to identify viable binders.

Core Features & Use Cases

• Diffusion-model based binder design generator with Protenix and AF2-IG confidence filtering
• Hotspot-guided design, adjustable binder length, and ranking across targets
• Structure validation and scoring for candidate prioritization in therapeutic protein engineering
• GPU-accelerated Docker workflow suitable for multi-target design campaigns

Quick Start

Use a GPU-enabled host with the PXDesign Docker image to run a full design pipeline. Example: docker run --rm --gpus all
-v $(pwd):/data
ghcr.io/open-prophetdb/pxdesign:arm64-blackwell
pxdesign pipeline --preset extended
-i /data/design.yaml
-o /data/output
--N_sample 100