NovaLocal Technical Requirements
NovaLocal Technical Requirements
NovaLocal enables protein structure prediction and modeling workflows to run locally through Docker Desktop integration in Protean 3D. System requirements vary depending on the workflows being used, with advanced structure prediction applications such as Boltz-2 requiring additional compute and GPU resources.
Review the requirements below to determine the recommended configuration for your environment.
Minimum Requirements for All NovaLocal Prediction Types
| Component | Requirement |
|---|---|
| Operating System | Windows 11 or supported Mac systems |
| Docker Platform | Current version of Docker Desktop must be running during prediction |
| Processor | Multi-core Intel or AMD processor |
| RAM | 16 GB minimum |
| Storage | 100 GB available disk space |
Additional Requirements for Machine Running a Boltz Structure Prediction
| Component | Recommended |
|---|---|
| GPU | RTX 3090 TI, RTX 4090, or RTX 5080 or 5090 GPU card |
| VRAM | 24 GB minimum; larger predictions require more |
| RAM | 64 GB |
| Storage | 1-4 TB available disk space |
| MMSeqs2 server access | Required for fully-local Boltz; see table below |
Additional Requirements for Machine Running MMSeqs2 Server
(needed only for fully-local Boltz)
| Component | Requirement |
|---|---|
| GPU | RTX 3090 TI, RTX 4090, or RTX 5080 or 5090 GPU card |
| RAM | 128 GB |
| Storage | 1 TB available disk space |
Compatibility Notes
- Boltz workflows currently require NVIDIA GPU acceleration.
- Mac systems are supported for select NovaLocal workflows, but GPU-accelerated Boltz predictions are not currently supported on macOS.
- Boltz workflows require access to an mmseqs alignment server, which may be hosted by DNASTAR, ColabFold, or configured locally.
Supported Methods in NovaLocal:
- AlphaFold-Multimer – Multi-chain protein structure prediction
- Boltz – Biomolecular complex structure prediction
- NovaFold Antibody – Antibody and antibody fragment modeling
- NovaDock – Protein-protein docking