---
name: alphafold3-protein-stability-protocol
description: Analyze protein stability and aggregation propensity for disease-related misfolding proteins using AlphaFold 3 predictions combined with stability scoring algorithms.
allowed-tools: WebFetch, Bash(python *), Bash(mkdir *), Bash(cp *), Bash(ls *), Bash(jq *), Bash(cd *)
---

# AlphaFold 3 Protein Stability & Misfolding Predictor Protocol

## Purpose

Analyze protein stability and predict aggregation or misfolding propensity for disease-related proteins. This workflow combines AlphaFold 3 structure predictions with stability scoring to identify vulnerable regions and aggregation-prone sequences, supporting research on amyloid diseases and proteinopathies.

## Inputs

Create an `inputs/` directory containing:

- `inputs/wildtype.json`: AlphaFold 3 JSON for the wild-type protein.
- `inputs/mutations.tsv` (optional): Known pathogenic mutations to analyze.
- `inputs/disease_context.md`: Disease name, known pathology, aggregation characteristics.
- `inputs/metadata.md`:
  - Protein name, UniProt ID
  - Normal function
  - Known instability or aggregation propensity
  - Literature on disease mechanism

## Pre-Run Checks

1. Confirm research use is permitted.
2. Validate protein sequence uses standard amino acid codes.
3. Verify disease context is documented.
4. Check for known aggregation-prone regions from literature.
5. Note that some proteins are intrinsically disordered (may not form stable structures).

## Step 1: Wild-Type Structure Prediction

### Route A: AlphaFold Server

1. Create job with wild-type protein.
2. Submit and download to `outputs/wildtype/`.

### Route B: Local AlphaFold 3

```bash
mkdir -p outputs/wildtype
python run_alphafold.py \
  --json_path=inputs/wildtype.json \
  --output_dir=outputs/wildtype
```

## Step 2: Analyze Structural Stability

Extract confidence and stability metrics:

```json
{
  "protein": "ALPHA_SYNUCLEIN",
  "source": "Homo sapiens",
  "length": 140,
  "pLDDT_mean": 45.3,
  "pLDDT_by_region": {
    "1-60": 32.1,
    "61-95": 78.5,
    "96-140": 28.4
  },
  "intrinsically_disordered_regions": [1, 2, 3, 4, 5, 96, 97, 98, 99, 100],
  "structured_domains": [61, 62, 63, 90, 91, 92],
  "predicted_stable_folds": [
    {"start": 61, "end": 95, "pLDDT": 78.5, "fold_type": "alpha-helical"}
  ],
  "stability_assessment": "partially_disordered"
}
```

## Step 3: Aggregation Propensity Analysis

Calculate aggregation scores using sequence-based predictors:

**TANGO (aggregation nucleation):**
- Sequence segments with TANGO > 5% indicate aggregation tendency.

**ZipperDB / 2UPR:**
- Energy < -23 kcal/mol indicates amyloid-forming segment.

** Waltz:**
- High-scoring regions indicate amyloid-forming potential.

**Simple hydrophobic analysis:**
```python
# Identify hydrophobic stretches > 5 residues
hydrophobic = "AVILMFYW"
sequence = "MVGGVV..."
# Windows of 5: count hydrophobic fraction
```

```json
{
  "aggregation_predictions": {
    "tango_regions": [
      {"start": 71, "end": 76, "sequence": "GAVVTG", "score": 12.5}
    ],
    "zipperdb_regions": [
      {"start": 68, "end": 72, "sequence": "GVGAV", "energy": -25.3}
    ],
    "hydrophobic_stretches": [
      {"start": 35, "end": 42, "hydrophobic_fraction": 0.75}
    ]
  },
  "aggregation_risk_assessment": "high",
  "key_amyloid_regions": ["68-72", "71-76"]
}
```

## Step 4: Mutation Impact Analysis

For each mutation:

1. Predict mutant structure or assess local stability change.
2. Evaluate effect on aggregation propensity.
3. Compare to wild-type.

```json
{
  "mutation": "A53T",
  "position": 53,
  "wildtype_residue": "A",
  "mutant_residue": "T",
  "local_pLDDT_wt": 32.5,
  "local_pLDDT_mut": 28.3,
  "aggregation_score_change": +15,
  "predicted_effect": "increased_aggregation",
  "literature": "known_parkinson_pathogenic"
}
```

## Step 5: Generate Stability Report

Write `outputs/stability_analysis.md`:

```markdown
# Protein Stability & Aggregation Propensity Analysis

## Protein
- Name: [name]
- UniProt ID: [ID]
- Source: [organism]
- Length: [N] residues
- Normal function: [description]

## Disease Context
- Associated disease: [name]
- Disease mechanism: [aggregation/misfolding/etc.]
- Affected tissues: [if known]

## Structure Prediction Quality

### Overall Assessment
- Mean pLDDT: [value]
- Confidence: [High/Medium/Low/Disordered]

### Disorder Profile
| Region | Residues | pLDDT | Assessment |
|--------|----------|-------|------------|
| N-terminal | 1-60 | [N] | [Disordered/Structured] |
| NAC region | 61-95 | [N] | [Disordered/Structured] |
| C-terminal | 96-140 | [N] | [Disordered/Structured] |

### Structured Domains
- Domain 1: residues [N-N], pLDDT [N]

## Aggregation Propensity Analysis

### Known Aggregation Regions
[From literature]
- Region 1: [positions], mechanism: [description]

### Predicted Aggregation Regions
| Region | Sequence | Predictor | Score | Risk |
|--------|----------|-----------|-------|------|
| [N-N] | [seq] | TANGO | [N] | High/Med/Low |

### Overall Aggregation Risk
- Assessment: [High/Medium/Low]
- Key regions: [list]

## Mutation Analysis (if applicable)

### [Mutation ID]: [mutation_string]
- Effect on structure: [stabilizing/destabilizing/no change]
- Effect on aggregation: [increased/decreased/no change]
- Correlation with disease: [supported/not supported/uncertain]

## Conformational Assessment

### Stable Regions
- Well-folded domains: [list]
- Stability confidence: [High/Medium/Low]

### Unstable/Disordered Regions
- Flexible regions: [list]
- Potential for disorder-to-order transition: [yes/no]

## Limitations
- AlphaFold 3 predictions may not capture:
  - Transient intermediates
  - Aggregation nucleus conformations
  - Membrane environment effects (for membrane proteins)
  - Cellular quality control factors
- Aggregation is kinetic, not purely thermodynamic
- Predictions do not account for:
  - pH, ionic strength, molecular crowding
  - Post-translational modifications
  - Proteostasis network effects
- Low-confidence regions may still form stable aggregates

## Recommendations
1. Validate predictions with ThT fluorescence assay
2. Test with transmission electron microscopy
3. Consider cellular models for aggregation kinetics
4. Test protective mutations for therapeutic potential
5. Use molecular dynamics for oligomerization simulation

## References
- AlphaFold 3: Abramson et al., Nature, 2024
- TANGO: Fernandez-Escamilla et al., Nat Biotech, 2004
- ZipperDB: Thompson et al., Structure, 2006
- Aggregation review: Chiti & Dobson, Annu Rev Biochem, 2017
```

## Success Criteria

- Wild-type structure is predicted and interpreted.
- Aggregation-prone regions are identified with supporting evidence.
- Mutations are analyzed for stability impact.
- Report connects structural features to disease mechanism.
- Limitations acknowledge the complexity of aggregation.

## Failure Modes

- Entire protein is disordered (low pLDDT throughout) → this may be expected for some proteins, note as intrinsically disordered
- No predicted aggregation regions → may be false negative, or protein requires triggers
- Mutation has no structural effect → effect may be kinetic/dynamic, not captured

## References

- AlphaFold 3: Abramson et al., Nature, 2024
- Aggregation prediction: TANGO, Fernandez-Escamilla et al., Nat Biotech, 2004
- Amyloid structure: Greenwald & Riek, Structure, 2020
- Protein misfolding: Chiti & Dobson, Annu Rev Biochem, 2017