Mitochondrial fatty acid β-oxidation

Reviewing the GO annotations of the whole enzyme spiral in human, fly and mouse

AI Gene Review · projects/FATTY_ACID_BETA_OXIDATION · 2026

Bottom line

  • β-oxidation strips two carbons per turn through four steps, each run by chain-length-specific enzymes.
  • We reviewed 713 GO annotations on 27 genes (10 human, 16 fly, mouse LCAD) and built a cross-species module.
  • 389 accepted, 237 non-core, 23 over-annotated, 30 modified, 14 removed, 20 undecided. Six blinded OpenScientist runs agreed with the reviews.

Who runs each step, by chain length

Why this pathway

  • One short, well-understood cycle that still concentrates five recurring curation problems:
    • chain-length specificity: use the specific MF where GO has one (VLCAD GO:0017099, MCAD GO:0070991, SCAD GO:0016937)
    • cross-gene transfer: SOAT cholesterol-esterification rows on the ACAT1 thiolase
    • moonlighting: HADH inhibits GLUD1; HADHA remodels cardiolipin
    • organelle: mitochondrion vs peroxisome
    • GO↔RHEA mapping: GO:0004300 maps to the (3E) reaction, not the (2E) crotonase

What a review looks like

ACADVL review page. The IBA very-long-chain ACAD activity is accepted; substrate binding is kept as non-core. Six IEA/ISS lipid-regulation rows transferred from mouse Acadvl (the true ortholog, not LCAD) are left undecided pending the mouse knockout evidence.

Results

Set Genes Ann. ACCEPT Non-core Over-ann. MODIFY REMOVE UNDECIDED
Human core spiral 10 442 251 124 10 24 14 19
Fly spiral + scully 11 196 92 88 9 6 0 1
Fly auxiliary isomerases 5 33 24 8 1 0 0 0
Mouse Acadl (LCAD) 1 42 22 17 3 0 0 0

Counts from the review YAMLs. Human UNDECIDED: ACADVL 6, ACAT1 6, ACADM 4, ACAD9 2, ACADS 1; fly: Egm. No NEW annotations.

Blinded checks agree with the reviews

Question OpenScientist verdict
fly Acat1 peroxisomal? Refuted: -EKL is not a PTS1; LOPIT says mitochondrion
fly Mtpalpha peroxisomal? Supported: fly-specific SKL PTS1, isoform-dependent targeting
ACAD9 very-long-chain? No: Thr-139/Ala-143 channel fits C16–C18 → long-chain
fly CG4860 short-chain? Over-specific: pocket Leu→Thr → general ACAD
fly Echs1, Mcad substrate range Conserved pockets; human-like range
fly step ③ = scully? Yes: 100% of 11 catalytic residues conserved

The cross-species module

modules/fatty_acid_beta_oxidation.yaml grounds each step in a human and a fly enzyme; its QC reports every grounded gene reviewed.

Status and next steps

  • ✅ Human spiral (10), fly spiral + auxiliary enzymes (16), mouse Acadl; module built; Reactome cross-check done.
  • ⬜ Remaining mouse orthologs (Acadvl, Acadm, Hadha, …), then rat and worm.
  • ⬜ Fly DECR1: no ortholog assignable from NCBI, UniProt, Ensembl or OrthoDB.
  • ⬜ FlyBase has taken one annotation from PMID:40519079; Arc42/CG4860 candidates listed.
  • ⬜ Schema: no way to negate an existing positive annotation (CG4860 case).

Read more: projects/FATTY_ACID_BETA_OXIDATION.md · modules/fatty_acid_beta_oxidation.yaml