NVIDIA NIM for OpenFold3

OpenFold3

• Overview
  • Key Features
  • Tooling
  • Advantages of NIMs
• Release Notes for NVIDIA NIM for OpenFold3
  • Release 1.3.0
    • Summary
    • Key Features
  • Release 1.2.0
    • Summary
    • Key Features
    • Notes and Limitations
  • Release 1.1.0
    • Summary
    • New Features
      • Structural Template Support
      • Additional GPU Support
      • Telemetry Control
    • Supported Features
  • Release 1.0.0
    • Summary
    • Supported Features
      • Entity Types
      • Multiple Entities and Complexes
      • Multiple Sequence Alignment (MSA) Support
      • Inference Parameters
      • Structure Templates
      • Output Formats
• Prerequisites
  • Known issues
  • NGC (NVIDIA GPU Cloud) Account
  • NGC CLI Tool
  • Set up your NIM cache
  • Installing the NVIDIA Driver, CUDA, Docker, and NVIDIA Container Toolkit Stack
    • Collect System Information
    • Installation Instructions by Architecture
  • Installation for amd64 / x86_64 Systems
    • 1. Find and Download Driver Package
    • 2. Direct Driver URLs
    • 3. Check and Purge Old Drivers (Optional but Recommended)
    • 4. Download the Driver
    • 5. Install the Local Repository
    • 6. Update Package Lists and Install Driver
    • 7. Reboot System
    • 8. Verify Driver Installation
    • 9. Install Docker
    • 10. Install NVIDIA Container Toolkit
    • 11. Verify the Complete Stack
  • Installation for arm64 / aarch64 DGX Systems
    • 1. Check NVIDIA Driver State
    • 2. Confirm OS Sees NVIDIA GPUs
    • 3. Verify System Requirements
    • 4. Update Linux Kernel Version (If Needed)
    • 5. Remove NVIDIA Libraries to Avoid Conflicts
    • 6. Download Package Repositories and Install DGX Tools
    • 7. Install GPU Driver
    • 8. Verify Driver Installation
    • 9. Install Docker and NVIDIA Container Toolkit
    • 10. Enable Docker for Non-Root User (Optional)
    • 11. Verify Complete Stack
  • Troubleshooting
    • Common Issues
    • Architecture-Specific Troubleshooting
    • Getting Additional Help
• Supported Hardware
  • GPU
  • Storage
  • CPU and System Memory
• Getting Started
  • Pull and run the NIM
  • Submit Health-Check Requests to the NIM as it Starts Up
  • Run Inference
• Example Requests
  • Using the OpenFold3 NIM
  • Try It Out
  • Try Out Other Prediction Use Cases
    • Protein-DNA Complex
    • Protein-Ligand with SMILES
    • Protein-Ligand with CCD Code
    • Protein with Structural Templates
    • Multi-Chain Protein Complex with MSAs
  • Understanding the API
    • Endpoint
    • Request Structure
    • Molecule Specification
    • MSA Structure
    • Structural Templates
    • Preparing MSAs
    • Response Structure
    • Field Reference Tables
      • Request-Level Fields
      • Input-Level Fields
      • Molecule Fields
      • AlignmentFileRecord Fields
      • StructuralTemplate Fields
    • Important Validation Rules
• Template Processing
  • Overview
  • How Template Processing Works
    • Chain Extraction
    • Sequence Alignment
    • Chain Scoring
    • Template Selection
    • Structure Guidance
  • Best Practices
    • When to Use Templates
    • Template Selection
    • Multi-Chain CIF Files
  • Template Requirements
  • Configuration
  • Related Documentation
• Performance in OpenFold3 NIM
  • NIM Accuracy
  • Factors Affecting NIM Performance
    • Hardware Factors
    • Input Complexity
    • Model Configuration
  • Performance Characteristics
    • Typical Runtimes
  • Performance Results on H100 (NVIDIA H100 80GB)
    • Structure Prediction Performance
    • Performance Analysis
    • Configuration
  • Performance Optimization Tips
    • Backend Selection
    • General Optimization Tips
  • Reproducing Performance Benchmarks
    • Overview
    • Prerequisites
      • Run the OpenFold3 NIM
      • Install the OpenStructure Docker Image
      • Download Reference Structures
    • Preparing Input JSON Files
      • Example Input JSON Structure
      • Create Input JSON Files
    • Benchmarking Script
      • Running a Single Benchmark
    • Understanding lDDT Metric
    • Running Benchmarks for Performance Table
      • Step 1: Prepare the Python scripts
      • Step 2: Create the benchmark workflow script
      • Step 3: Run the complete workflow
    • Expected Output Format
  • Troubleshooting
    • General Performance Issues
    • Backend-Specific Issues
    • Benchmarking Issues
• Configuration Alternatives at Runtime
  • Start the OpenFold3 NIM
  • Using an Alternative Port for OpenFold3 NIM Requests
  • Backend Optimization Options
    • Running with PyTorch+cuEquivariance
    • Running with PyTorch+DeepSpeed
  • Template Configuration Options
    • Configuring Template Chain Selection Threshold
• Observability
  • About Observability
  • Health Check Endpoints
    • Readiness Check
    • Liveness Check
  • Metrics
    • Available Metrics Summary
  • Collecting Metrics with OpenTelemetry Collector
    • Prerequisites
    • OpenTelemetry Collector Configuration
    • Checking for Port Conflicts
    • Alternative Configuration for Port Conflicts
    • Running the Collector
  • Related Information
• API Reference
  • OpenAPI Schema
    • Reference
• EULA