Document dGPredictor KEGG carry-over defect in the NAR 2026 Methods drafts

[freiburgermsu]

Summary

Expands the multi-source thermodynamics Methods draft (Papers/NAR_Update_2026/drafts/methods_multi_source_thermodynamics.md) and its inlined copy in MANUSCRIPT.md with the discovery and correction of a systematic identifier carry-over defect in the original (KEGG-mediated) dGPredictor integration.

The finding being documented

• Of the 27,715 reactions carrying a stored dGPredictor record, 17,271 (62%) hold an estimate computed for a KEGG reaction id that is not among the reaction's aliases (the reaction records and Unique_ModelSEED_Reaction_Aliases.txt agree on all 17,271 cases).
• Root cause: a stale-variable bug in the 2023 staging notebook — the KEGG-alias extraction loop never reset its holding variable between reactions, so every KEGG-less reaction silently inherited the id of the nearest preceding KEGG-bearing reaction (e.g. rxn00019, a nitroalkane oxidase, carries RuBisCO's R00024 energy of −8.63 ± 0.04 kcal/mol; rxn13478–rxn13481 all inherited R09245 from rxn13477).
• A database-wide fingerprint test explains 100.0% of the 17,784 contaminated staged entries by carry-over, with zero exceptions.
• The contaminated subset shows null correlation with eQuilibrator and Group Contribution (r = −0.001 / −0.006) versus r = 0.61 / 0.74 for the correctly-mapped subset — explaining the method's previously puzzling low agreement with other sources.
• Because Update_Reaction_dGPredictor_Energies.py never validates staged KEGG ids against aliases, every Rerun_Thermodynamics.sh pass re-propagated the contamination, and the promotion policy's lowest-uncertainty rule favored the deceptively tight copied errors: 7,466 contaminated values were promoted into canonical deltag/deltagerr/reversibility.

What this PR changes

Documentation only — no data or script changes:

• New Methods subsection (defect finding → forensic confirmation → why it persisted → correction path), styled to mirror the existing GC Convention-A subsection.
• The retrained dGPredictor-ModelSEED source is noted as structurally immune (keyed directly by ModelSEED rxn id, no KEGG detour) and unaffected.
• [TBD] markers on the correction pieces not yet landed: regenerated staging JSONs (corrected extraction script lives in the dGPredictor repo), an alias-validation guard in the ingest script, and the re-promotion pass repairing the 7,466 canonical triples. The draft's "Open loose ends" list tracks these, plus a suggested regression test pinning known carry-over cases.
• Results [NUMBERS PENDING] stub now flags that the head-to-head ΔrG′ comparison must be computed from post-repair records.

🤖 Generated with Claude Code

https://claude.ai/code/session_01XoJyv1ukSyPfME8WaHewJh