AffiniBind scans your protein's surface, binder design engines work in parallel, and quality filters screen every candidate before it reaches the lab. The goal is predicted binding proteins that actually work.
engine: affinibind-v0.5.0 · containerized
Many protein-design workflows end after candidate generation. AffiniBind extends beyond it: each stage produces a data contract that feeds into the next, creating an auditable chain of evidence.
WeaveSeq runs a vertically integrated pipeline from target preparation through results export. Each stage produces a schema-validated Parquet table consumed by the next: no silent fallbacks, no heuristic conversion, and complete traceability from raw structure to ranked candidate shortlist.
The target protein structure is cleaned up: non-standard pieces removed, missing atoms filled in, and surface properties calculated. This creates a clean foundation for everything that follows.
Four AI models independently scan the protein surface for promising binding regions. ScanNet and MPBind use deep learning to spot binding-prone areas. MaSIF analyzes the 3D surface shape. APBS maps the electrostatic charge. Their predictions are merged into a single confidence score for each surface location.
High-confidence surface spots are grouped into contiguous patches using spatial clustering. Duplicate patches are merged away, and the remaining patches are ranked. An interactive 3D viewer shows where each patch sits on the protein surface.
Three protein design engines run simultaneously on cloud GPUs, each using a different approach to create candidate binders. FreeBindCraft iteratively refines designs guided by structure prediction. PXDesign generates backbones shaped to fit the target interface. BoltzGen builds protein backbones using a diffusion-based method. Jobs are distributed across GPU instances and managed automatically.
Every candidate is scored across twelve developability criteria that predict real-world behavior. Solubility estimates whether the protein will express well. 3D-SAP checks for sticky patches that cause aggregation. PROPKA computes charge state and stability. PRODIGY predicts binding strength (ΔG, Kd). SSIPe evaluates how mutations affect binding energy. Additional screens cover flexibility, packing quality, and other manufacturing risks.
All quality metrics are combined into a single developability score. Candidates are grouped by similarity to ensure diverse options for testing, and the best representative from each group is selected. The final ranking balances binding strength, solubility, stability, and ease of manufacturing.
Everything is packaged into ready-to-use formats: interactive HTML reports, spreadsheets, and structured data files. All intermediate results are saved for traceability. The output feeds directly into the next stage of the lab pipeline: initial screening, protein production, and functional testing.
All candidates are scored through a shared ranking layer covering binding confidence, interface quality, predicted affinity, developability, and manufacturing feasibility. Representative figures from a recent campaign are shown below. See the glossary for metric definitions.
Integrate AffiniBind into your discovery pipeline and work directly with our engineering team, from target analysis through assay-ready reagents.