| name | ipsae |
| description | Binder design ranking using ipSAE (interprotein Score from Aligned Errors). Use this skill when: (1) Ranking binder designs for experimental testing, (2) Filtering BindCraft or BoltzGen outputs, (3) Comparing AF2/AF3/Boltz predictions, (4) Predicting binding success rates, (5) Need better ranking than ipTM or iPAE.
For structure prediction, use chai or alphafold. For QC thresholds, use protein-qc.
|
| license | MIT |
| category | evaluation |
| tags | ["ranking","scoring","binding"] |
| source | https://github.com/adaptyvbio/protein-design-skills |
ipSAE Binder Ranking
Overview
ipSAE outperforms ipTM and iPAE for binder design ranking with 1.4x higher precision.
Paper: What's wrong with AlphaFold's ipTM score
Installation
git clone https://github.com/DunbrackLab/IPSAE.git
cd IPSAE && pip install numpy
How to run
AlphaFold2
python ipsae.py scores_rank_001.json unrelaxed_rank_001.pdb 15 15
AlphaFold3
python ipsae.py fold_model_full_data_0.json fold_model_0.cif 10 10
Boltz1
python ipsae.py pae_model_0.npz model_0.cif 10 10
Key parameters
| Parameter | Recommended | Description |
|---|
| PAE cutoff | 10-15 | Threshold for contacts |
| Distance cutoff | 10-15 | Max CA-CA distance (Å) |
Output format
Chain-pair scores (_chains.csv):
chain_A,chain_B,ipSAE_min,pDockQ,LIS,n_contacts
A,B,0.72,0.65,0.45,42
Recommended thresholds
| Metric | Standard | Stringent |
|---|
| ipSAE_min | > 0.61 | > 0.70 |
| LIS | > 0.35 | > 0.45 |
| pDockQ | > 0.5 | > 0.6 |
Decision tree
Should I use ipSAE?
├─ Designed binders → ipSAE ✓ (1.4x better than ipTM)
├─ Natural complexes → ipTM is fine
├─ Different length constructs → ipSAE ✓
└─ Single proteins → Not applicable
Troubleshooting
| Error | Cause | Fix |
|---|
KeyError: 'pae' | Wrong PAE format | Check if AF2/AF3/Boltz format |
ValueError: no contacts | No interface | Check chain IDs, reduce cutoffs |
Next: Select top designs (ipSAE_min > 0.61) → experimental validation.
