easyFold

easyFold is an interactive, structure-aware platform for AlphaFold3 job management, visualization, and domain-level interpretation.
It integrates confidence metrics (pLDDT, PAE), contact maps, and automated domain segmentation to enable deep, interpretable analysis of predicted protein structures.

🔹 AlphaFold3 Job Management

• Docker-based AlphaFold3 execution
• User-level job submission, tracking, and result management
• Administrator dashboard for global job monitoring and control
• Unique job IDs to avoid conflicts across users

🔹 Interactive Structure Visualization

• 3D structure visualization using Mol*
• Supports PDB / mmCIF outputs
• Optional domain-colored structures (via B-factor encoding)

🔹 Confidence & Contact Analysis

• pLDDT per-residue curve
• PAE heatmap visualization
• Contact map (CA–CA, configurable cutoff)
• Mouse-based region selection on PAE and contact maps

🔹 Domain-Aware Interpretation (Optional)

• Integrated Merizo for automatic domain segmentation
• Domain boundary visualization:
  • Sequence domain bar
  • Domain overlays on PAE and contact maps
  • Structure highlighting by domain
• Quantitative intra- / inter-domain contact density analysis

🔹 Linked, Interpretable Views

• Domain bar ↔ PAE ↔ contact map ↔ 3D structure are fully linked
• Selecting a region or domain highlights corresponding residues across views
• Designed for structure-informed domain interpretation, not just visualization

---

📦 Installation

requirement first Python3.10+

1. Clone repository

git clone https://github.com/your-org/easyFold.git
cd easyFold

2. Create virtual environment

python3 -m venv .venv
source .venv/bin/activate
pip3 install -r requirements.txt

3. Configure Docker

1. Docker installed and running
2. AlphaFold3 image available (e.g. cford38/alphafold3)
3. GPU support recommended (--gpus all)

4. Start server

python3 -m uvicorn app:app --reload

Then open: 👉 http://127.0.0.1:8000

👤 User Roles

Regular User

Submit AlphaFold3 jobs

Check jobs before submission

View and download results

Explore structure, confidence, contact maps, and domains

🧠 Design Philosophy

easyFold is not just a wrapper for AlphaFold.

It is designed to support:

• Domain-level reasoning
• Structure-aware interpretation
• Confidence-guided analysis
• Exploration of inter-domain coupling and organization

This makes easyFold suitable for:

• Multi-domain proteins
• Large bacterial proteins
• Toxin systems
• Structure-based functional annotation studies

---

Configure host paths (input/output/models/AFDB)

Monitor CPU / GPU / memory usage

View and manage all users’ jobs

Stop or delete jobs

Configure execution limits (single-node mode)

📊 Result Dashboard

Each job provides a multi-tab dashboard:

Overview: job metadata and artifacts

Structure: interactive 3D visualization

Confidence: pLDDT, PAE, contact map

Domains: Merizo-based domain segmentation and statistics

Compare: multi-model / seed comparison

Logs: real-time execution logs

🧪 Domain & Contact Metrics

easyFold automatically computes:

Intra-domain contact density

Inter-domain contact density

Observed vs. possible contact ratios

These metrics support:

Domain validity assessment

Structural coupling analysis

Domain-level functional hypotheses

📖 Citation

If you use easyFold in your research, please cite:

Li, J. et al. easyFold: an interactive platform for structure-aware domain interpretation of AlphaFold predictions. Manuscript in preparation.

📬 Contact

Maintained by: Jinhui Li For issues, suggestions, or collaboration, please open an issue or contact the author.