Skip to main content
Ctrl+K
RFD4-ProteinaRFD4-Proteina
Home
Quickstart Overview Abstract Latent reasoning In silico results Wet lab Code & weights Citation
Latent space Architecture Benchmarks Documentation
  • Quickstart
  • Installation
  • Examples
  • How-to guides
  • Concepts
  • Reference
  • Contributing
  • GitHub
  • Quickstart
  • Installation
  • Examples
  • How-to guides
  • Concepts
  • Reference
  • Contributing
  • GitHub

Section Navigation

  • Workflow
    • Annotated CIF
    • Conditioning
    • Partial diffusion and inverse folding
    • Metric guidance
    • Reward guided search
    • Metrics
    • Refolding
  • Concepts

Concepts#

How RFD4-Proteina’s pieces fit together: the input format, the ways to run the model, and how search and guidance steer it. For step-by-step tasks, start at Quickstart instead; for lookup tables and syntax, see Reference.

Workflow The full pipeline as an interactive diagram, from annotated CIF through generation, search, refolding, and metrics. Annotated CIF What the format is, why it keeps structure and intent in one file, and how conditioning attaches at different granularities. Conditioning Why conditions behave the way they do: active masks, conditions that imply other conditions, and why charges and bonds matter. Partial diffusion and inverse folding Start denoising from a structure you already have instead of from noise. Metric guidance Differentiate a scalar objective and steer the sampling trajectory toward it. Reward guided search Branch generation at checkpoints and keep the candidates that score best. Metrics What each metric family measures, how binder and target selections resolve, and where a run can stop early. Refolding Why an independent structure prediction model checks each design, and how a design can carry more than one refold.

previous

FAQ

next

Workflow

© Copyright 2026, Baker Lab and NVIDIA GenAIR.

Created using Sphinx 8.2.3.

Built with the PyData Sphinx Theme 0.23.0.