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)