🧊 AIPC: Tomography Alignment & Spike Analysis

Cryo-electron tomography (cryo-ET) and subtomogram averaging pipeline for reconstructing and analyzing viral spike structures.

---

📌 Table of Contents

• Overview
• Objectives
• Workflow
• Key Concepts
• Results
• Project Structure
• Tools & Technologies
• Figures
• Conclusion
• Future Work

---

🔬 Overview

This project focuses on cryo-electron tomography (cryo-ET) and subtomogram averaging to reconstruct and analyze viral spike structures.

The workflow covers the full pipeline from raw tilt-series data to a refined 3D model of viral spikes:

• Tilt-series alignment
• Tomogram reconstruction
• Particle picking
• Subtomogram averaging
• Structural refinement

---

🎯 Objectives

• Align tilt-series data with high accuracy
• Reduce residual alignment error
• Isolate individual viral spikes
• Perform subtomogram averaging
• Generate a high-quality 3D model

---

⚙️ Workflow

Raw Tilt-Series Data
        │
        ▼
┌───────────────────────┐
│ 1. Data Preprocessing │──→ Binning (factor 2 → 4)
│    (Binning)          │    Improved SNR, reduced size
└───────────┬───────────┘
            │
            ▼
┌───────────────────────┐
│ 2. Tilt-Series        │──→ Patch tracking alignment
│    Alignment          │    800×800 patches, 0.33 overlap
│                       │    Target residual: 0.35 nm
└───────────┬───────────┘
            │
            ▼
┌───────────────────────┐
│ 3. Tomogram           │──→ 3D volume from aligned tilt-series
│    Reconstruction     │    Full specimen thickness (~300 nm)
└───────────┬───────────┘
            │
            ▼
┌───────────────────────┐
│ 4. Particle Picking   │──→ Manual spike identification
│    (Spike Extraction) │    Clean, isolated subtomograms
└───────────┬───────────┘
            │
            ▼
┌───────────────────────┐
│ 5. Subtomogram        │──→ Cylindrical mask (r=20, h=35)
│    Averaging          │    Centering shift (10 units)
└───────────┬───────────┘
            │
            ▼
┌───────────────────────┐
│ 6. Angular Constraints│──→ C3 symmetry (trimer)
│    & Symmetry         │    Azimuth: 120°, Cone: 60°
└───────────┬───────────┘
            │
            ▼
┌───────────────────────┐
│ 7. Refinement &       │──→ Iterative refinement
│    Visualization      │    Final model in Chimera
└───────────────────────┘
            │
            ▼
    ✅ 3D Spike Structure

---

1. Data Preprocessing (Binning)

• Applied binning (factor 2 → 4) to reduce data size
• Improved Signal-to-Noise Ratio (SNR)
• Trade-off: reduced resolution for faster processing

---

2. Tilt-Series Alignment

Used patch tracking for alignment with optimized parameters:

Parameter	Value
Patch size	800 × 800
Overlap	0.33
Iterations	Increased for convergence
Target residual error	0.35 nm

---

3. Tomogram Reconstruction

• Generated 3D tomogram from aligned tilt-series
• Included full specimen thickness (~300 nm)

---

4. Particle Picking (Spike Extraction)

• Identified viral spikes manually
• Avoided selecting multiple spikes per particle
• Ensured clean and isolated subtomograms

---

5. Subtomogram Averaging

Applied cylindrical mask for focused averaging:

Parameter	Value
Mask shape	Cylindrical
Radius	20
Height	35
Centering shift	10 units

---

6. Angular Constraints & Symmetry

• Assumed C3 symmetry (trimer structure)
• Reduced azimuth rotation range to 120°
• Used cone aperture: 60°

---

7. Refinement & Visualization

• Iterative refinement of spike structure
• Final model visualized in UCSF Chimera

---

📊 Key Concepts

🔹 Binning

Advantage	Disadvantage
✅ Faster processing	❌ Reduced resolution
✅ Improved SNR
✅ Smaller data size

---

🔹 Template Matching vs Patch Tracking

Method	Best For
Patch tracking	General alignment of tilt-series
Template matching	Spherical features (e.g., gold beads)

---

🔹 Subtomogram Challenges

• Irregular spike distribution on viral surface
• Risk of averaging multiple spikes → blurred results
• Requires careful manual picking for clean isolation

---

📈 Results

Metric	Status
Reduced residual alignment error	✅
Improved particle alignment	✅
Clean subtomogram averages	✅
High-quality 3D spike reconstruction	✅

Successfully reconstructed two 3D models (3d_0-1, 3d_0-2) with improved alignment consistency from additional views.

---

📂 Project Structure

aipc-tomography/
│
├── report/                 # Full project report
│   └── cryo_et_report.pdf
│
├── scripts/                # Workflow notes and commands
│   └── pipeline_notes.txt
│
├── parameters/             # Alignment and averaging settings
│   └── alignment_params.txt
│
├── results/                # Summary of results
│   └── results_summary.txt
│
├── figures/                # Screenshots and visualizations
│   ├── tilt_series/        # Tilt-series alignment (fiducials)
│   ├── tomogram/           # 3D tomogram slices
│   ├── picking/            # Particle picking screenshots
│   ├── averages/           # Masked subtomogram averages
│   └── chimera/            # Final Chimera model renders
│
└── README.md               # Project documentation

---

🛠️ Tools & Technologies

Category	Tools
Alignment & Reconstruction	IMOD (newstack, alignment tools)
Processing Pipeline	Cryo-ET processing tools
Visualization	UCSF Chimera

---

📸 Figures

• Tilt-series alignment (fiducials)
• 3D tomogram slices
• Particle picking visualization
• Masked subtomogram averages
• Final Chimera model

---

✅ Conclusion

This project demonstrates how careful parameter tuning, masking, and symmetry constraints significantly improve subtomogram averaging and 3D reconstruction quality in cryo-ET.

---

🔮 Future Work

• Automated particle picking using deep learning
• Higher resolution refinement
• Advanced alignment methods
• Integration with RELION / Dynamo / EMAN2