second_mut/README.md

# 🏆 SPICE second_mut — Second survivor batch (acidic collapse, pH 2.0)

> **Milestone date: 2026-08-14**
> Second independent survivor batch from the clean pipeline — this time rescuing 7QF3
> (miniSOG) in an **acidic collapse environment (pH 2.0 / 330 K)**. Same Q quality as the
> first batch (`first_mut`), but a distinct rescue strategy.
> (Produced after the Head B' alias + `_sane_ca` local-clash guard fix restored mutant
> construction via the wild-type-backbone fallback.)

---

## Protein

- **7QF3** = miniSOG (R57Q mutant), *Arabidopsis thaliana*, flavoprotein photosensitizer (same as first_mut)
- 116 residues (the modeled chain)

## Rescue environment (Env_fail)

- **pH 2.0 / 330 K / ionic 0** — strongly acidic stress
- 7QF3 collapses here; the mutants below survive the full 20-step window here

## Five surviving mutants (episode 1)

| File | Mutations | Q | steps | Strategy |
|---|---|---|---|---|
| `pseudo_7qf3_0_20.npz` | 50:Q>K | 0.92 | 20 | aggressive |
| `pseudo_7qf3_1_20.npz` | 4:S>Y; 49:D>K | 0.90 | 20 | aggressive |
| `pseudo_7qf3_2_20.npz` | 4:S>F; 36:L>W; 49:D>K | 0.89 | 20 | aggressive |
| `pseudo_7qf3_3_20.npz` | 3:K>Y | 0.92 | 20 | aggressive |
| `pseudo_7qf3_4_20.npz` | 47:E>K | 0.91 | 20 | aggressive |

- **Q-gate threshold 0.5; all scored 0.89–0.92** → genuine fold retention

## Mutation pattern (tentative — hypothesis, not proven)

- **K additions cluster at 47/49/50 (4/5)** — the N-terminal cap region of the main helix
  (residues 50–59). Lys is helix-N-cap-favorable → tentative hypothesis: *stabilizing the
  helix N-cap under acid-induced unfolding* (at pH 2, protonated Asp/Glu disrupt helix capping).
- **Aromatic additions at 3/4/36 (4/5)** (Y/F/W) — packing restoration.
- Note: positional convergence is **weaker** than `first_mut` (no position hits 3+/5);
  the strategy-level convergence (K + aromatic) is real, but the exact mechanism is a
  hypothesis pending collapse-mechanism analysis.

## Structure quality (verified 2026-08-14, `analysis_metrics.csv`)

| Metric | Value | Reading |
|---|---|---|
| Rg | 14.44–14.56 Å | correct compact size for 116 aa ✅ |
| Adjacent Cα bond | 3.95–3.96 Å | proper chain geometry ✅ |
| Helix content | 23.3–27.6% | plausible α-helical level ✅ |
| Nearest non-adjacent pair | 4.11–4.41 Å | no clashes ✅ |
| Q (re-derived vs log) | 0.89–0.92, **exact match** | genuine fold retention ✅ |

> Q recomputed from archived coordinates (run's native-contact definition, ref = 7QF3 Cα,
> cutoff 8 Å) matches the run log exactly.

## Honest caveats

- Q validates **structural** retention, not **function** (cofactor binding / photosensitizer activity unverified)
- The helix-N-cap mechanism is a **hypothesis**; it needs collapse-mechanism analysis (where does pH-2 unfolding start?) and single-mutant controls
- This is the **third collapse environment** (pH 7.5–8 `first_mut`, pH 10 alkaline, pH 2 acidic) — strengthens the "loop adapts the rescue chemistry to the environment" claim

## Paper placement

- Candidate: third-environment evidence in §3.3 ("strategy adapts per environment"), pending the full run + possible additional survivors
- Reference: same ref 21 (Lafaye 2022)

## Files

```
second_mut/
├── README.md                    # this milestone record
├── analysis_metrics.csv         # structure-quality metrics (uniform method)
├── run_log.txt                  # raw HPC run-log excerpt (acidic episode)
├── pseudo_7qf3_{0..4}_20.npz    # 5 pseudo-labels (seq + env + coords)
```