[edkerk]

Closes #880

QC

• New report qc_split_compartments.csv: reactions whose chemistry spans more than one compartment, ignoring the metabolites they transport. There are 37 on develop and 29 here. Some are correct (e.g. HGSNAT, GPD2), so only a rise is flagged.
• The annotation-consistency check now also flags a metabolite whose ID suffix differs from its compartment. This found the cause of Cytosolic metabolites in mitochondrial non-transport reactions #880: MAM00208m had compartment c, now m. The remaining case, MAM20076n, is fixed by fix: correct two metabolite compartments and curate the NAAA ceramidase GPRs #1088.

Curation

• MAR06779: product (3S)-phytanic acid instead of dibenzo[a,l]pyrene-11,12-diol.
• MAR01392: fully cytosolic, subsystem "Prostaglandin biosynthesis".
• MAR00081, MAR08317: GPR B4GALT1–4 instead of B3GNT3. MAR08317's product is moved to the Golgi, as the new MAM02186g.
• Duplicate metabolites merged: MAM01915 → MAM01884, MAM03231e → MAM03232e.
• Deprecated as duplicates: MAR08279, MAR01340, MAR08300, MAR09333, MAR10470, MAR10473.
• MAM02186: removed annotations that describe 4-isopropylaniline.

Growth is unchanged.

[github-actions]

Model quality report

⚠️ 5 pre-existing finding(s), no regressions vs develop. Non-blocking. Gates: duplicate keys 0 ✅, growth 1.9 ✅.

Section	Status
Model & network checks (19)	⚠️ 5 pre-existing
Model file & metabolic tasks (5)	✅ all pass
MEMOTE	✅ 63.8% (core subset)
Full report	model_qc_summary.md

Improved:

• Reactions flagged by MACAW dead-end test: 1106
• Mass-imbalanced reactions: 69

Pre-existing:

• Reactions split across compartments: 26
• Cross-refs inconsistent across compartments: 2
• Reactions flagged as MACAW duplicates: 342
• Charge-imbalanced reactions: 196
• Structure vs formula/charge inconsistencies: 306

✅ unchanged · ✨ improved vs develop · ⚠️ pre-existing, non-blocking · ❌ regression

Full workflow run · this comment is edited as results come in

[edkerk]

/run gene-essentiality

[github-actions]

Gene essentiality was run on this pull request via GH Actions.

Gene essentiality: effect of this change

Checked 9 gene(s) directly affected plus 1986 more that share a metabolite with one of them.

A gene counts as changed below only if the flip agrees in direction across at least 3 of the 5 cell-line models; fewer than that is grouped as likely noise from ftINIT's run-to-run variability instead.

Category	Status
Growth effect changed (vs Hart 2015)	ℹ️ 1 not scored by Hart
Other role changed (not Hart-comparable)	✅ 0
Likely noise (<3/5 lines)	ℹ️ 11
No change	✅ 1983

Full detail: gene-essential_summary.md.

✅ unchanged · ✨ correct vs Hart 2015 · ❌ wrong vs Hart 2015 · ℹ️ not comparable to Hart 2015

The unscoped, every-gene matrix behind the table above is committed to gene-essential.csv.

Note: each run posts its results in place of its own start notice; earlier results link here.

[edkerk]

The one gene flagged in the gene-essentiality run, HSD17B7 (3/5 lines, not scored by Hart 2015), is an unstable gene rather than an effect of this PR. None of the reactions changed here shares a metabolite with an HSD17B7 reaction (other than cofactors), and its growth ratio flips between runs of an unchanged model: HeLa and RPE1 were 0 in the #1044 run, 0.99 in the #1094 run and 0 again here; DLD1 and GBM were already 0 before this PR.

[edkerk]

Corrections in #1105 after an evidence review:

Change	Model entities	Evidence
I-antigen and iso-nLc8Cer are the same compound: MAM03584g is deprecated into MAM02186g, which gets its BiGG/Recon3D id, MetaNetX id and structure. MAR00082 and MAR01083 now use MAM02186g. The constructed HMR id m02186g is removed	MAM03584g, MAM02186c/g, MAR00082, MAR01083	KEGG G00078 is named "iso-nLc8Cer; LacNAc-Lc6Cer; I-antigen", with the same structure as MAM03584; ModelSEED cpd21564 lists I-antigen as a synonym
MAR08317 was then identical to MAR00082 and is deprecated into it (MAR00082 keeps HMR_8317 and RCR20529, rxnRetired B3GNT314g;HMR_8317;MAR08317). After this PR MAM02186g had no consumer, so MAR08317 could carry no flux	MAR08317 -> MAR00082	Same MetaNetX (MNXR96181) and BiGG (B3GNT314g) ids
MAR00082 transfers galactose, not GlcNAc: GPR B3GNT3 -> B4GALT1, B4GALT2, B4GALT3 or B4GALT4 (as on MAR00081); name "beta-1,4-galactosyltransferase (G00077)"; references PMID:10072769 (a cDNA signal-anchor screen) -> PMID:9405390, PMID:9792633	MAR00082	B3GNT3 has only UDP-GlcNAc activities (UniProt Q9Y2A9). B4GALT3 makes nLc6Cer (PMID:9405390); B4GALT4 galactosylates glycolipids with a terminal beta-GlcNAc (PMID:9792633)
The PTGS peroxidase steps were charge-imbalanced (2 H+ and no reductant). Now PGG + NADPH + H+ -> PGH + NADP+ + H2O, irreversible, in all three copies	MAR01392 [c], MAR01393 [r], MAR01307 [r]	Rhea RHEA:50432 and RHEA:42600 (PGG + AH2 = PGH + A + H2O); NADPH as reductant follows the model's lumped PTGS reactions (MAR08557, MAR08558, MAR02581)

Not changed: PTGS1/2 are ER-membrane enzymes, but only the cytosolic PG1 branch connects to PGE1 (PTGES3, MAR01394); the ER copy (MAR01391/MAR01393) is the blocked one. Choosing one compartment for the COX steps is an open point in #1109.