ARBA00028655 phosphatidylinositol biosynthetic process (GO:0006661)

View original ARBA rule on UniProt

Type: ARBA
Status: COMPLETE
Action: SPLIT
Confidence: 0.95

Description

Assigns GO:0006661 "phosphatidylinositol biosynthetic process" to proteins matching any of 25 alternative condition sets built from InterPro entries, CATH FunFams and taxon constraints. Only two of the 25 sets (CS18, CS19) identify the enzyme that actually makes phosphatidylinositol - CDP-diacylglycerol--inositol 3-phosphatidyltransferase (CDIPT/PIS, EC 2.7.8.11). The remaining 23 sets capture phosphoinositide kinases, phosphoinositide phosphatases, non-catalytic PI3K/VPS34 regulatory subunits, GPI-anchor pathway enzymes, DPM1, and two promiscuous structural folds - all of which act on, consume or are merely adjacent to PI rather than synthesising it.

Analysis Summary

Condition-set counts describe the sets recorded in this review, which may omit the full rule.

0
Domain Pairs Analyzed
25
Recorded condition sets
0
Subset Relationships
0
Redundant Annotations

Review Summary

ARBA00028655 is a wrong-branch, near-total-false-positive rule. GO:0006661 is effectively a leaf term denoting formation of the unphosphorylated phosphatidylinositol molecule itself (CDP-diacylglycerol + myo-inositol -> PI + CMP, EC 2.7.8.11). Only condition sets 18 and 19 (CATH FunFams 1.20.120.1760:FF:000003 in Vertebrata and :FF:000021 in Fungi, both CDIPT/PIS) identify that enzyme. The other 23 condition sets identify PI3K/PI4K/PIP5K/PIP4K catalytic subunits (11 sets), phosphoinositide phosphatases (4 sets), GPI-anchor and dolichol-phosphate-mannose pathway enzymes (5 sets: PIGN, PIGL, PIGO, PGAP5/MPPE1, DPM1), and two non-catalytic regulatory subunits (PIK3R1/p85-alpha, UVRAG). Crucially, the phosphoinositide branch is a SIBLING of GO:0006661, not a descendant: the QuickGO ancestor closure of GO:0046854 "phosphatidylinositol phosphate biosynthetic process" (checked 2026-08-15) does not contain GO:0006661 - the two terms first meet at GO:0046474 glycerophospholipid biosynthetic process. So these are not over-broad annotations a curator could refine downward; they are simply in the wrong subtree. A QuickGO census on 2026-08-15 found 4,782 GO:0006661 / ECO:0000256 / GO_REF:0000117 annotations attributed to ARBA00028655; in a stratified sample of 1,000 accessions only 8 (0.8%) were CDIPT/PIS, while 63.8% were myotubularin-family phosphoinositide 3-phosphatases and one hit was cytochrome b-c1 complex subunit 9 (UQCR10), a pure domain-promiscuity artifact. In the human subset (169 annotations, enumerated exhaustively) only 3 (1.8%) are PI synthase. Taxon constraints are biologically unmotivated (Homo-only INPPL1, Mus-only PIP5K1C, Catarrhini-only "Kinase"), and eight sets - including the DPM1 set that produced the S. pombe dpm1 annotation queried in geneontology/go-annotation#5835 - carry no taxon constraint at all. The rule is also incomplete in the one direction that would help: no condition set targets CDS1/CDS2, the other family legitimately part of de novo PI biosynthesis.

Action Rationale

SPLIT rather than DEPRECATE, because condition sets 18 and 19 are correct and worth keeping: CDIPT/PIS is precisely the family GO:0006661 exists to describe, and a PIS-only rule would be a good rule. Everything else should be dissolved into the branch it actually belongs to - GO:0046854 (or a product-specific child such as GO:0036092) for the kinases, GO:0046856 phosphatidylinositol dephosphorylation for the phosphatases, GO:0006506 GPI anchor biosynthetic process for the GPI enzymes, and GO:0180047 dolichol phosphate mannose biosynthetic process for DPM1 - with the non-catalytic (CS15 PIK3R1/p85, CS20 UVRAG) and fold-promiscuous (CS8 PH-domain superfamily, CS21 Zn-finger RING/FYVE/PHD superfamily) sets deleted outright rather than retargeted. If the maintainers prefer not to split, DEPRECATE is the correct fallback, since retaining the rule in its current form means retaining ~4,740 wrong-branch annotations to salvage ~40 correct ones. The ~4,740 existing IEA annotations not derived from CS18/CS19 should be retracted. The confidence value below is confidence in this recommendation.

GO Annotations

GO:0006661 - phosphatidylinositol biosynthetic process
Aspect: P

Rule Definition

Condition Sets

Condition Set 1

3 condition(s)
Notes:

INCORRECT. IPR000387 (PTP-like domain) + IPR010569 (myotubularin phosphatase domain) identifies the myotubularin family (MTM1, MTMR1-4, MTMR7). These are phosphoinositide 3-phosphatases: PI3P -> PI and PI(3,5)P2 -> PI5P. PI is a dephosphorylation product of a turnover reaction, not a de novo biosynthetic step; the deep research calls this verdict 'Outright incorrect'. Better BP: GO:0046856 phosphatidylinositol dephosphorylation. The Primates taxon restriction is an annotation-bias artifact - myotubularins are pan-eukaryotic.

Condition Set 2

2 condition(s)
Notes:

INCORRECT. IPR002013 (SAC domain) captures SACM1L/Sac1, SYNJ1/SYNJ2, FIG4 and INPP5F - all phosphoinositide phosphatases (e.g. Sac1: PI4P -> PI). Same dephosphorylation-is-not-biosynthesis error as CS1. This condition set is a major contributor to the observed false positives (SAC-domain proteins dominate the human hit set). Better BP: GO:0046856. Haplorrhini restriction is arbitrary.

Condition Set 3

2 condition(s)
Notes:

INCORRECT. PIK3C3/VPS34 class III PI3K FunFam. VPS34 phosphorylates pre-existing PI to PI3P; it consumes PI rather than making it. Better BP: GO:0036092 phosphatidylinositol-3-phosphate biosynthetic process (or GO:0046854 phosphatidylinositol phosphate biosynthetic process), which is a SIBLING of GO:0006661, not a descendant.

Condition Set 4

3 condition(s)
Notes:

INCORRECT. Three FunFams of the class IA PI3K catalytic subunit (PIK3CA/PIK3CB p110alpha/beta). These make PI(3,4,5)P3 from PI(4,5)P2. Better BP: GO:0046854 phosphatidylinositol phosphate biosynthetic process.

Condition Set 5

3 condition(s)
Notes:

INCORRECT. Duplicate of the CS4 concept using the sibling FunFams for p110beta (1.10.1070.11:FF:000001, 1.25.40.70:FF:000004, 2.60.40.150:FF:000046). Same class IA PI3K biology, same wrong term, and largely redundant with CS4. Better BP: GO:0046854.

Condition Set 6

3 condition(s)
Notes:

INCORRECT. Class II PI3K PIK3C2B FunFams; produce PI3P/PI(3,4)P2 from pre-existing lipids. Better BP: GO:0046854 (or GO:0036092).

Condition Set 7

3 condition(s)
Notes:

INCORRECT. Type II PIP4K (PIP4K2B) FunFams: PI5P -> PI(4,5)P2. Substrate is already a phosphoinositide. Better BP: GO:0046854. Primates restriction is unmotivated.

Condition Set 8

2 condition(s)
Notes:

INCORRECT AND PROMISCUOUS. 2.30.29.30 is the PH-domain-like superfamily; FF:000038 carries the misleading label 'Myotubularin 1, isoform CRA_a' but in practice matches unrelated PH-domain proteins - PLEKHA3/FAPP1, PLEKHA8/FAPP2, PITPNM1 and PITPNM2 all appear in the emitted annotation set. A bare PH-domain FunFam cannot support any catalytic BP term. This condition set should be deleted outright rather than retargeted.

Condition Set 9

2 condition(s)
Notes:

INCORRECT. FunFam 3.30.470.160:FF:000001 is labelled only 'Kinase' (PIP-kinase insert region), restricted to Catarrhini. An uninformative label plus a primate-only taxon is not a defensible basis for a metabolic BP annotation. Better BP if retained: GO:0046854; preferably delete.

Condition Set 10

2 condition(s)
Notes:

INCORRECT. PIP5K1C (type I PIP5K) FunFam restricted to Mus. PI4P -> PI(4,5)P2. Better BP: GO:0046854. The genus-level Mus restriction is an annotation-bias artifact.

Condition Set 11

1 condition(s)
Notes:

INCORRECT. FunFam 3.30.800.10:FF:000009 'Phosphatidylinositol 4-phosphate 5-kinase its3' (S. pombe Its3). Its3 makes PI(4,5)P2 from PI4P; the S. pombe PI synthase is a distinct gene (pis1). No taxon restriction at all, so this set fires very broadly. Better BP: GO:0046854.

Condition Set 12

1 condition(s)
Notes:

INCORRECT. PIGN (GPI ethanolamine phosphate transferase 1). Acts on GPI intermediates that already contain PI; it does not make PI. Better BP: GO:0006506 GPI anchor biosynthetic process. No taxon restriction.

Condition Set 13

3 condition(s)
Notes:

INCORRECT. Class IB PI3K PIK3CG (p110gamma) FunFams; PI(4,5)P2 -> PI(3,4,5)P3. Better BP: GO:0046854. Also largely redundant with CS4/CS5.

Condition Set 14

3 condition(s)
Notes:

INCORRECT. Class II PIK3C2G FunFams restricted to Glires. Better BP: GO:0046854. Redundant in concept with CS6.

Condition Set 15

3 condition(s)
Notes:

INCORRECT - NON-CATALYTIC. PIK3R1/p85alpha PI3K REGULATORY subunit FunFams (Rodentia). p85 has no lipid-kinase activity; it stabilises and restrains p110. No PI or PIP biosynthetic BP term is appropriate. If a term is wanted, use the MF GO:0035014 phosphatidylinositol 3-kinase regulator activity, not a BP biosynthesis term. Delete from this rule.

Condition Set 16

3 condition(s)
Notes:

INCORRECT. INPPL1/INPP5D (SHIP) 5-phosphatase FunFams restricted to Homo. PI(3,4,5)P3 -> PI(3,4)P2, a catabolic/turnover reaction. Better BP: GO:0046856 phosphatidylinositol dephosphorylation. A genus-Homo restriction on a conserved enzyme family is an annotation-bias artifact.

Condition Set 17

2 condition(s)
Notes:

INCORRECT. FunFam 1.10.1070.11:FF:000019 is labelled 'Phosphatidylinositol 4-kinase beta 1' and restricted to Viridiplantae; this pairing is coherent, since plants have genuine PI4-kinase beta enzymes (Arabidopsis PI4Kbeta1 Q9FMJ0/At5g64070 and PI4Kbeta2 Q0WPX9/At5g09350, both with that recommended name) and CATH FunFam names take a representative member. The set is wrong for the same reason every other kinase set here is wrong: PI4K makes PI4P from PI, which is GO:0046854, not the formation of PI itself. Better BP: GO:0046854.

Condition Set 18

2 condition(s)
Notes:

CORRECT - RETAIN. FunFam 1.20.120.1760:FF:000003 'CDP-diacylglycerol--inositol 3-phosphatidyltransferase' (CDIPT/PIS, EC 2.7.8.11) in Vertebrata. This is the only enzyme that actually forms the PI molecule (CDP-DAG + myo-inositol -> PI + CMP). GO:0006661 is exactly right here.

Condition Set 19

2 condition(s)
Notes:

CORRECT - RETAIN. FunFam 1.20.120.1760:FF:000021 'CDP-diacylglycerol--inositol 3-phosphatidyltransferase' (PIS) in Fungi. Same reaction as CS18. GO:0006661 is correct. Note the CDP-DAG/PIS route is also present outside Eukaryota (e.g. mycobacterial PgsA1), so a eukaryote-only framing of the rule as a whole would be too narrow.

Condition Set 20

2 condition(s)
Notes:

INCORRECT - NON-CATALYTIC. UVRAG FunFam (Euteleostomi). UVRAG is a Beclin-1-binding regulatory/targeting subunit of VPS34 complex II; VPS34 is the catalytic kinase. UVRAG neither makes nor phosphorylates PI. No biosynthetic BP term applies; macroautophagy/endosomal transport terms describe it. Delete from this rule.

Condition Set 21

2 condition(s)
Notes:

INCORRECT AND PROMISCUOUS. FunFam 3.30.40.10:FF:000073 sits in the Zn-finger RING/FYVE/PHD superfamily and is labelled 'myotubularin-related protein 4 isoform X2' (Mammalia). Even taking the label at face value, MTMR4 is a phosphoinositide 3-phosphatase (better BP: GO:0046856); the underlying superfamily is a promiscuous small Zn-binding fold and is unsafe as a sole condition.

Condition Set 22

2 condition(s)
Notes:

INCORRECT. PIGL (N-acetylglucosaminyl-phosphatidylinositol de-N-acetylase, GlcNAc-PI -> GlcN-PI), Eutheria. Downstream of PI, consumes it. Better BP: GO:0006506 GPI anchor biosynthetic process.

Condition Set 23

2 condition(s)
Notes:

INCORRECT. PIGO (GPI ethanolamine phosphate transferase 3), Euarchontoglires. Modifies GPI precursors. Better BP: GO:0006506.

Condition Set 24

1 condition(s)
Notes:

INCORRECT. PGAP5/MPPE1 metallophosphoesterase; removes ethanolamine phosphate from Man2 during GPI remodelling. Better BP: GO:0006506 GPI anchor biosynthetic process (or a GPI-remodelling term). No taxon restriction.

Condition Set 25

1 condition(s)
Notes:

INCORRECT - THE SET FLAGGED BY CURATORS. FunFam 3.90.550.10:FF:000036 'Dolichol-phosphate mannosyltransferase subunit 1' (DPM1). DPM1 makes Dol-P-Man from GDP-mannose and dolichol-phosphate; it has no role in forming PI. This is the route by which S. pombe dpm1 (SPAC31G5.16c, UniProt O14466) received GO:0006661, the annotation questioned in geneontology/go-annotation issue #5835. Correct BP: GO:0180047 dolichol phosphate mannose biosynthetic process (already annotated for O14466), with GO:0006506 by pathway participation. The deep research states this is 'not supportable'.

Assessments

OVERLY_COMPLEX

25 condition sets for a single, effectively-leaf GO term is far beyond the point at which a rule can be reasoned about, and the complexity is not the good kind: it does not represent genuine biological diversity among PI-synthesising enzymes, it represents the accretion of every domain family that happens to co-occur with the string "phosphatidylinositol" in protein names. The one biologically coherent concept in the rule - PI synthase - is covered by just two sets (CS18, CS19), and those two are near-duplicates differing only in taxon (Vertebrata vs Fungi) and could be merged. Meanwhile the class IA/IB PI3K catalytic subunit is expressed four separate times (CS4, CS5, CS13, and partly CS3) and the class II PI3K twice (CS6, CS14), each as a different trio of FunFams from the same three CATH superfamilies (1.10.1070.11, 1.25.40.70, 2.60.40.150). A parsimonious version of the correct part of this rule would be a single condition set on the CDIPT/PIS FunFam or on the corresponding InterPro PIS family entry.

CONTRADICTED

The commissioned deep research (Falcon / Edison Scientific Literature, run 2026-08-15) directly contradicts the rule for 23 of its 25 condition sets. Its executive conclusion is that GO:0006661 is not a correct blanket annotation for these families, and its per-family verdict table assigns "Outright incorrect" - explicitly defined there as a chemically different process rather than merely a less precise term - to every PI3K/PI4K/PIP5K/PIP4K family, every phosphoinositide phosphatase family (myotubularins, SACM1L/Sac1, synaptojanins, INPP5D), every GPI-pathway enzyme (PIGL, PIGN/PIGO, PGAP5/MPPE1), DPM1, UVRAG and PIK3R1/p85. The only families it endorses are CDIPT/PIS (the direct PI-forming enzyme, matching CS18 and CS19) and CDS1/CDS2 as precursor suppliers - and CDS1/CDS2 are not among the rule's conditions. The report is explicit that annotating DPM1 to GO:0006661 is not supportable, which is exactly the case raised by curators in geneontology/go-annotation#5835. Note two honest limits of this evidence: the report found no retrievable GO Consortium adjudication of ARBA00028655 itself, so no formal Consortium ruling should be claimed; and it refers to GO:0046854 by its superseded label "phosphatidylinositol phosphorylation" (the current primary label, verified in QuickGO on 2026-08-15, is "phosphatidylinositol phosphate biosynthetic process") - a labelling slip that does not affect its reasoning. Underlying primary and review literature (Dickson & Hille 2019; Blunsom & Cockcroft 2020; Fox et al. 2020; Cao et al. 2008; Kinoshita 2020; Imbach et al. 2000) is consistent with the report.

Supporting Evidence:

  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: GO:0006661 is not a biochemically correct blanket annotation for the listed families
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: PIS/CDIPT is the enzyme that directly creates the PI molecule
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: Only CDS1/CDS2 and CDIPT/PIS warrant GO:0006661 among the assessed catalytic families; CDIPT/PIS is the only direct PI-forming enzyme.
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: Assigning GO:0006661 directly to these proteins conflates
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: Kinases add phosphate to the 3-, 4-, or 5-hydroxyl of the inositol headgroup
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: MTM1/MTMR2 use PI3P and PI(3,5)P2 as substrates, yielding PI and PI5P, respectively.
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: A terminal product being PI does not convert a salvage or catabolic reaction into the canonical biosynthetic pathway.
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: None produces the PI lipid moiety.
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: Annotating DPM1 to GO:0006661 is not supportable.
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: VPS34 is the catalytic lipid kinase and major producer of PI3P; UVRAG is a targeting/regulatory component, not a phosphoinositide-metabolizing enzyme.
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: p85 has no PI/PIP lipid-kinase catalytic activity.
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: established molecular function lies in a chemically different process, not merely that a more specific child term is preferable.
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: Retain GO:0006661 only for condition sets that identify
  • file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md: I found no retrievable, authoritative GO Consortium issue or publication explicitly documenting a discussion of
SIGNIFICANT

Two distinct overlap problems. (1) Genuine near-duplication of the same biological concept across sets: CS4, CS5 and CS13 each specify a trio of FunFams drawn from the same three CATH superfamilies (1.10.1070.11 kinase catalytic, 1.25.40.70 helical/PIK, 2.60.40.150 C2) for class IA/IB PI3K catalytic subunits, and CS6/CS14 do the same for class II PI3K; the single superfamily 1.10.1070.11 appears in seven different condition sets (CS3, CS4, CS5, CS6, CS13, CS14, CS17), 2.60.40.150 in five (CS4, CS5, CS6, CS13, CS20) and 3.30.800.10 in three (CS7 PIP4K2B, CS10 PIP5K1C, CS11 Its3). CS18/CS19 are the same FunFam superfamily 1.20.120.1760 split only by taxon. (2) Overlap in the emitted protein set even where the conditions are formally disjoint: CS1 (IPR000387 + IPR010569) and CS2 (IPR002013 SAC domain) both hit synaptojanins, which carry a SAC domain alongside a 5-phosphatase domain, and CS8's PH-domain FunFam overlaps the lipid-transfer proteins also reachable through other sets. Because every set emits the same single GO term, this redundancy does not change the annotation content - it just multiplies the number of routes by which a wrong annotation can be produced and makes the rule harder to repair incrementally. Note that no automated pairwise_overlap statistics were computed for this review (the `just analyze-rule` step was not run for this rule), so the overlaps described here are structural/biological rather than quantitative.

MISMATCHED

Not merely too broad - wrong branch. GO:0006661 is defined as "The chemical reactions and pathways resulting in the formation of phosphatidylinositol, any glycophospholipid in which the sn-glycerol 3-phosphate residue is esterified to the 1-hydroxyl group of 1D-myo-inositol", and it is effectively a leaf (its only children are the regulation terms). It therefore asserts specifically that the protein helps form the unphosphorylated PI molecule. The products of the PI kinases fall under GO:0046854 phosphatidylinositol phosphate biosynthetic process, whose QuickGO ancestor closure (is_a + part_of, checked 2026-08-15) is GO:0046474, GO:0046486, GO:0006650, GO:0006644, GO:0008654, GO:0045017, GO:0008610, GO:0006629, GO:0019637, GO:0006793, GO:0090407, GO:0009058, GO:0044238, GO:0008152, GO:0009987, GO:0008150 - GO:0006661 is absent, and the two terms first meet at GO:0046474 glycerophospholipid biosynthetic process. Phosphoinositide kinase annotations under this rule are therefore in a sibling subtree, not a descendant one, and cannot be repaired by refining the term downward. The same argument applies to the phosphatases (whose process is GO:0046856 phosphatidylinositol dephosphorylation, under the dephosphorylation/catabolism branch), to the GPI enzymes (GO:0006506) and to DPM1 (GO:0180047). For CS18/CS19 alone the term is exactly right. All replacement term labels were verified against QuickGO on 2026-08-15.

TOO_NARROW

The taxon constraints are internally incoherent rather than uniformly wrong, so no single enum value fits cleanly; TOO_NARROW is recorded because the dominant, systematic defect is that pan-eukaryotic enzyme families are pinned to arbitrary narrow clades. Across the 25 sets the constraints are Primates (CS1, CS7), Haplorrhini (CS2, CS8), Catarrhini (CS9), Homo (CS16), Mus (CS10), Glires (CS14), Rodentia (CS15), Eutheria (CS22), Euarchontoglires (CS23), Mammalia (CS21), Vertebrata (CS18), Euteleostomi (CS20), Viridiplantae (CS17), Fungi (CS19), Eukaryota (CS3) - and none at all for CS4, CS5, CS6, CS11, CS12, CS13, CS24 and CS25. Myotubularins, SAC-domain phosphatases, PI3Ks, PIP5Ks, the GPI pathway and DPM1 are all pan-eukaryotic, so a Homo-only INPPL1 set or a Mus-only PIP5K1C set reflects which clade happened to be densely annotated in the training data, not where the biology exists. Conversely the eight unconstrained sets fire across all of UniProt - which is precisely how S. pombe dpm1 (SPAC31G5.16c) acquired GO:0006661 via CS25 and triggered geneontology/go-annotation#5835. The Viridiplantae restriction on CS17 is a genuine exception to this pattern and is not counted as an artifact here: plants do have PI4-kinase beta enzymes, so that constraint matches the family it names. Even the two correct sets are narrower than the biology: the CDP-DAG/PIS route is conserved across eukaryotes and PI synthase activity is not eukaryote-exclusive, so Vertebrata (CS18) and Fungi (CS19) could reasonably be relaxed to Eukaryota or removed.

References (8)

Raw YAML

View Source YAML
id: ARBA00028655
description: >-
  Assigns GO:0006661 "phosphatidylinositol biosynthetic process" to proteins matching any of
  25 alternative condition sets built from InterPro entries, CATH FunFams and taxon
  constraints. Only two of the 25 sets (CS18, CS19) identify the enzyme that actually makes
  phosphatidylinositol - CDP-diacylglycerol--inositol 3-phosphatidyltransferase (CDIPT/PIS,
  EC 2.7.8.11). The remaining 23 sets capture phosphoinositide kinases, phosphoinositide
  phosphatases, non-catalytic PI3K/VPS34 regulatory subunits, GPI-anchor pathway enzymes,
  DPM1, and two promiscuous structural folds - all of which act on, consume or are merely
  adjacent to PI rather than synthesising it.
status: COMPLETE
rule_type: ARBA
rule:
  rule_id: ARBA00028655
  condition_sets:
  - number: 1
    conditions:
    - condition_type: INTERPRO
      value: IPR000387
      curie: InterPro:IPR000387
      label: "Tyrosine-specific protein phosphatases domain"
      interpro_type: DOMAIN
      negated: false
    - condition_type: INTERPRO
      value: IPR010569
      curie: InterPro:IPR010569
      label: "Myotubularin-like, phosphatase domain"
      interpro_type: DOMAIN
      negated: false
    - condition_type: TAXON
      value: Primates
      curie: NCBITaxon:9443
      label: "Primates"
      negated: false
    notes: "INCORRECT. IPR000387 (PTP-like domain) + IPR010569 (myotubularin phosphatase domain) identifies the myotubularin family (MTM1, MTMR1-4, MTMR7). These are phosphoinositide 3-phosphatases: PI3P -> PI and PI(3,5)P2 -> PI5P. PI is a dephosphorylation product of a turnover reaction, not a de novo biosynthetic step; the deep research calls this verdict 'Outright incorrect'. Better BP: GO:0046856 phosphatidylinositol dephosphorylation. The Primates taxon restriction is an annotation-bias artifact - myotubularins are pan-eukaryotic."
  - number: 2
    conditions:
    - condition_type: INTERPRO
      value: IPR002013
      curie: InterPro:IPR002013
      label: "SAC domain"
      interpro_type: DOMAIN
      negated: false
    - condition_type: TAXON
      value: Haplorrhini
      curie: NCBITaxon:376913
      label: "Haplorrhini"
      negated: false
    notes: "INCORRECT. IPR002013 (SAC domain) captures SACM1L/Sac1, SYNJ1/SYNJ2, FIG4 and INPP5F - all phosphoinositide phosphatases (e.g. Sac1: PI4P -> PI). Same dephosphorylation-is-not-biosynthesis error as CS1. This condition set is a major contributor to the observed false positives (SAC-domain proteins dominate the human hit set). Better BP: GO:0046856. Haplorrhini restriction is arbitrary."
  - number: 3
    conditions:
    - condition_type: FUNFAM
      value: 1.10.1070.11:FF:000002
      curie: CATH.FunFam:1.10.1070.11:FF:000002
      label: "Phosphatidylinositol 3-kinase catalytic subunit type 3"
      negated: false
    - condition_type: TAXON
      value: Eukaryota
      curie: NCBITaxon:2759
      label: "Eukaryota"
      negated: false
    notes: "INCORRECT. PIK3C3/VPS34 class III PI3K FunFam. VPS34 phosphorylates pre-existing PI to PI3P; it consumes PI rather than making it. Better BP: GO:0036092 phosphatidylinositol-3-phosphate biosynthetic process (or GO:0046854 phosphatidylinositol phosphate biosynthetic process), which is a SIBLING of GO:0006661, not a descendant."
  - number: 4
    conditions:
    - condition_type: FUNFAM
      value: 1.10.1070.11:FF:000006
      curie: CATH.FunFam:1.10.1070.11:FF:000006
      label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
      negated: false
    - condition_type: FUNFAM
      value: 1.25.40.70:FF:000001
      curie: CATH.FunFam:1.25.40.70:FF:000001
      label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
      negated: false
    - condition_type: FUNFAM
      value: 2.60.40.150:FF:000041
      curie: CATH.FunFam:2.60.40.150:FF:000041
      label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
      negated: false
    notes: "INCORRECT. Three FunFams of the class IA PI3K catalytic subunit (PIK3CA/PIK3CB p110alpha/beta). These make PI(3,4,5)P3 from PI(4,5)P2. Better BP: GO:0046854 phosphatidylinositol phosphate biosynthetic process."
  - number: 5
    conditions:
    - condition_type: FUNFAM
      value: 1.10.1070.11:FF:000001
      curie: CATH.FunFam:1.10.1070.11:FF:000001
      label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
      negated: false
    - condition_type: FUNFAM
      value: 1.25.40.70:FF:000004
      curie: CATH.FunFam:1.25.40.70:FF:000004
      label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit beta"
      negated: false
    - condition_type: FUNFAM
      value: 2.60.40.150:FF:000046
      curie: CATH.FunFam:2.60.40.150:FF:000046
      label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
      negated: false
    notes: "INCORRECT. Duplicate of the CS4 concept using the sibling FunFams for p110beta (1.10.1070.11:FF:000001, 1.25.40.70:FF:000004, 2.60.40.150:FF:000046). Same class IA PI3K biology, same wrong term, and largely redundant with CS4. Better BP: GO:0046854."
  - number: 6
    conditions:
    - condition_type: FUNFAM
      value: 1.10.1070.11:FF:000003
      curie: CATH.FunFam:1.10.1070.11:FF:000003
      label: "Phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit beta"
      negated: false
    - condition_type: FUNFAM
      value: 2.60.40.150:FF:000036
      curie: CATH.FunFam:2.60.40.150:FF:000036
      label: "phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit beta"
      negated: false
    - condition_type: FUNFAM
      value: 3.30.1010.10:FF:000001
      curie: CATH.FunFam:3.30.1010.10:FF:000001
      label: "Phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit beta"
      negated: false
    notes: "INCORRECT. Class II PI3K PIK3C2B FunFams; produce PI3P/PI(3,4)P2 from pre-existing lipids. Better BP: GO:0046854 (or GO:0036092)."
  - number: 7
    conditions:
    - condition_type: FUNFAM
      value: 3.30.800.10:FF:000002
      curie: CATH.FunFam:3.30.800.10:FF:000002
      label: "Phosphatidylinositol 5-phosphate 4-kinase type-2 beta"
      negated: false
    - condition_type: FUNFAM
      value: 3.30.810.10:FF:000003
      curie: CATH.FunFam:3.30.810.10:FF:000003
      label: "Phosphatidylinositol 5-phosphate 4-kinase type-2 beta"
      negated: false
    - condition_type: TAXON
      value: Primates
      curie: NCBITaxon:9443
      label: "Primates"
      negated: false
    notes: "INCORRECT. Type II PIP4K (PIP4K2B) FunFams: PI5P -> PI(4,5)P2. Substrate is already a phosphoinositide. Better BP: GO:0046854. Primates restriction is unmotivated."
  - number: 8
    conditions:
    - condition_type: FUNFAM
      value: 2.30.29.30:FF:000038
      curie: CATH.FunFam:2.30.29.30:FF:000038
      label: "Myotubularin 1, isoform CRA_a"
      negated: false
    - condition_type: TAXON
      value: Haplorrhini
      curie: NCBITaxon:376913
      label: "Haplorrhini"
      negated: false
    notes: "INCORRECT AND PROMISCUOUS. 2.30.29.30 is the PH-domain-like superfamily; FF:000038 carries the misleading label 'Myotubularin 1, isoform CRA_a' but in practice matches unrelated PH-domain proteins - PLEKHA3/FAPP1, PLEKHA8/FAPP2, PITPNM1 and PITPNM2 all appear in the emitted annotation set. A bare PH-domain FunFam cannot support any catalytic BP term. This condition set should be deleted outright rather than retargeted."
  - number: 9
    conditions:
    - condition_type: FUNFAM
      value: 3.30.470.160:FF:000001
      curie: CATH.FunFam:3.30.470.160:FF:000001
      label: "Kinase"
      negated: false
    - condition_type: TAXON
      value: Catarrhini
      curie: NCBITaxon:9526
      label: "Catarrhini"
      negated: false
    notes: "INCORRECT. FunFam 3.30.470.160:FF:000001 is labelled only 'Kinase' (PIP-kinase insert region), restricted to Catarrhini. An uninformative label plus a primate-only taxon is not a defensible basis for a metabolic BP annotation. Better BP if retained: GO:0046854; preferably delete."
  - number: 10
    conditions:
    - condition_type: FUNFAM
      value: 3.30.800.10:FF:000001
      curie: CATH.FunFam:3.30.800.10:FF:000001
      label: "phosphatidylinositol 4-phosphate 5-kinase type-1 gamma"
      negated: false
    - condition_type: TAXON
      value: Mus
      curie: NCBITaxon:10088
      label: "Mus"
      negated: false
    notes: "INCORRECT. PIP5K1C (type I PIP5K) FunFam restricted to Mus. PI4P -> PI(4,5)P2. Better BP: GO:0046854. The genus-level Mus restriction is an annotation-bias artifact."
  - number: 11
    conditions:
    - condition_type: FUNFAM
      value: 3.30.800.10:FF:000009
      curie: CATH.FunFam:3.30.800.10:FF:000009
      label: "Phosphatidylinositol 4-phosphate 5-kinase its3"
      negated: false
    notes: "INCORRECT. FunFam 3.30.800.10:FF:000009 'Phosphatidylinositol 4-phosphate 5-kinase its3' (S. pombe Its3). Its3 makes PI(4,5)P2 from PI4P; the S. pombe PI synthase is a distinct gene (pis1). No taxon restriction at all, so this set fires very broadly. Better BP: GO:0046854."
  - number: 12
    conditions:
    - condition_type: FUNFAM
      value: 3.40.720.10:FF:000015
      curie: CATH.FunFam:3.40.720.10:FF:000015
      label: "GPI ethanolamine phosphate transferase 1"
      negated: false
    notes: "INCORRECT. PIGN (GPI ethanolamine phosphate transferase 1). Acts on GPI intermediates that already contain PI; it does not make PI. Better BP: GO:0006506 GPI anchor biosynthetic process. No taxon restriction."
  - number: 13
    conditions:
    - condition_type: FUNFAM
      value: 1.10.1070.11:FF:000010
      curie: CATH.FunFam:1.10.1070.11:FF:000010
      label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit gamma isoform"
      negated: false
    - condition_type: FUNFAM
      value: 1.25.40.70:FF:000006
      curie: CATH.FunFam:1.25.40.70:FF:000006
      label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit gamma isoform"
      negated: false
    - condition_type: FUNFAM
      value: 2.60.40.150:FF:000087
      curie: CATH.FunFam:2.60.40.150:FF:000087
      label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit gamma isoform"
      negated: false
    notes: "INCORRECT. Class IB PI3K PIK3CG (p110gamma) FunFams; PI(4,5)P2 -> PI(3,4,5)P3. Better BP: GO:0046854. Also largely redundant with CS4/CS5."
  - number: 14
    conditions:
    - condition_type: FUNFAM
      value: 1.10.1070.11:FF:000013
      curie: CATH.FunFam:1.10.1070.11:FF:000013
      label: "Phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit gamma"
      negated: false
    - condition_type: FUNFAM
      value: 1.25.40.70:FF:000010
      curie: CATH.FunFam:1.25.40.70:FF:000010
      label: "Phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit gamma"
      negated: false
    - condition_type: TAXON
      value: Glires
      curie: NCBITaxon:314147
      label: "Glires"
      negated: false
    notes: "INCORRECT. Class II PIK3C2G FunFams restricted to Glires. Better BP: GO:0046854. Redundant in concept with CS6."
  - number: 15
    conditions:
    - condition_type: FUNFAM
      value: 1.10.287.1490:FF:000001
      curie: CATH.FunFam:1.10.287.1490:FF:000001
      label: "Putative phosphatidylinositol 3-kinase regulatory subunit alpha"
      negated: false
    - condition_type: FUNFAM
      value: 1.10.555.10:FF:000035
      curie: CATH.FunFam:1.10.555.10:FF:000035
      label: "Phosphatidylinositol 3-kinase regulatory subunit alpha"
      negated: false
    - condition_type: TAXON
      value: Rodentia
      curie: NCBITaxon:9989
      label: "Rodentia"
      negated: false
    notes: "INCORRECT - NON-CATALYTIC. PIK3R1/p85alpha PI3K REGULATORY subunit FunFams (Rodentia). p85 has no lipid-kinase activity; it stabilises and restrains p110. No PI or PIP biosynthetic BP term is appropriate. If a term is wanted, use the MF GO:0035014 phosphatidylinositol 3-kinase regulator activity, not a BP biosynthesis term. Delete from this rule."
  - number: 16
    conditions:
    - condition_type: FUNFAM
      value: 3.30.505.10:FF:000035
      curie: CATH.FunFam:3.30.505.10:FF:000035
      label: "phosphatidylinositol 3,4,5-trisphosphate 5-phosphatase 1"
      negated: false
    - condition_type: FUNFAM
      value: 3.60.10.10:FF:000005
      curie: CATH.FunFam:3.60.10.10:FF:000005
      label: "phosphatidylinositol 3,4,5-trisphosphate 5-phosphatase 1"
      negated: false
    - condition_type: TAXON
      value: Homo
      curie: NCBITaxon:9605
      label: "Homo"
      negated: false
    notes: "INCORRECT. INPPL1/INPP5D (SHIP) 5-phosphatase FunFams restricted to Homo. PI(3,4,5)P3 -> PI(3,4)P2, a catabolic/turnover reaction. Better BP: GO:0046856 phosphatidylinositol dephosphorylation. A genus-Homo restriction on a conserved enzyme family is an annotation-bias artifact."
  - number: 17
    conditions:
    - condition_type: FUNFAM
      value: 1.10.1070.11:FF:000019
      curie: CATH.FunFam:1.10.1070.11:FF:000019
      label: "Phosphatidylinositol 4-kinase beta 1"
      negated: false
    - condition_type: TAXON
      value: Viridiplantae
      curie: NCBITaxon:33090
      label: "Viridiplantae"
      negated: false
    notes: "INCORRECT. FunFam 1.10.1070.11:FF:000019 is labelled 'Phosphatidylinositol 4-kinase beta 1' and restricted to Viridiplantae; this pairing is coherent, since plants have genuine PI4-kinase beta enzymes (Arabidopsis PI4Kbeta1 Q9FMJ0/At5g64070 and PI4Kbeta2 Q0WPX9/At5g09350, both with that recommended name) and CATH FunFam names take a representative member. The set is wrong for the same reason every other kinase set here is wrong: PI4K makes PI4P from PI, which is GO:0046854, not the formation of PI itself. Better BP: GO:0046854."
  - number: 18
    conditions:
    - condition_type: FUNFAM
      value: 1.20.120.1760:FF:000003
      curie: CATH.FunFam:1.20.120.1760:FF:000003
      label: "CDP-diacylglycerol--inositol 3-phosphatidyltransferase"
      negated: false
    - condition_type: TAXON
      value: Vertebrata
      curie: NCBITaxon:7742
      label: "Vertebrata"
      negated: false
    notes: "CORRECT - RETAIN. FunFam 1.20.120.1760:FF:000003 'CDP-diacylglycerol--inositol 3-phosphatidyltransferase' (CDIPT/PIS, EC 2.7.8.11) in Vertebrata. This is the only enzyme that actually forms the PI molecule (CDP-DAG + myo-inositol -> PI + CMP). GO:0006661 is exactly right here."
  - number: 19
    conditions:
    - condition_type: FUNFAM
      value: 1.20.120.1760:FF:000021
      curie: CATH.FunFam:1.20.120.1760:FF:000021
      label: "CDP-diacylglycerol--inositol 3-phosphatidyltransferase"
      negated: false
    - condition_type: TAXON
      value: Fungi
      curie: NCBITaxon:4751
      label: "Fungi"
      negated: false
    notes: "CORRECT - RETAIN. FunFam 1.20.120.1760:FF:000021 'CDP-diacylglycerol--inositol 3-phosphatidyltransferase' (PIS) in Fungi. Same reaction as CS18. GO:0006661 is correct. Note the CDP-DAG/PIS route is also present outside Eukaryota (e.g. mycobacterial PgsA1), so a eukaryote-only framing of the rule as a whole would be too narrow."
  - number: 20
    conditions:
    - condition_type: FUNFAM
      value: 2.60.40.150:FF:000148
      curie: CATH.FunFam:2.60.40.150:FF:000148
      label: "UV radiation resistance associated gene"
      negated: false
    - condition_type: TAXON
      value: Euteleostomi
      curie: NCBITaxon:117571
      label: "Euteleostomi"
      negated: false
    notes: "INCORRECT - NON-CATALYTIC. UVRAG FunFam (Euteleostomi). UVRAG is a Beclin-1-binding regulatory/targeting subunit of VPS34 complex II; VPS34 is the catalytic kinase. UVRAG neither makes nor phosphorylates PI. No biosynthetic BP term applies; macroautophagy/endosomal transport terms describe it. Delete from this rule."
  - number: 21
    conditions:
    - condition_type: FUNFAM
      value: 3.30.40.10:FF:000073
      curie: CATH.FunFam:3.30.40.10:FF:000073
      label: "myotubularin-related protein 4 isoform X2"
      negated: false
    - condition_type: TAXON
      value: Mammalia
      curie: NCBITaxon:40674
      label: "Mammalia"
      negated: false
    notes: "INCORRECT AND PROMISCUOUS. FunFam 3.30.40.10:FF:000073 sits in the Zn-finger RING/FYVE/PHD superfamily and is labelled 'myotubularin-related protein 4 isoform X2' (Mammalia). Even taking the label at face value, MTMR4 is a phosphoinositide 3-phosphatase (better BP: GO:0046856); the underlying superfamily is a promiscuous small Zn-binding fold and is unsafe as a sole condition."
  - number: 22
    conditions:
    - condition_type: FUNFAM
      value: 3.40.50.10320:FF:000002
      curie: CATH.FunFam:3.40.50.10320:FF:000002
      label: "Probable N-acetylglucosaminyl-phosphatidylinositol de-N-acetylase"
      negated: false
    - condition_type: TAXON
      value: Eutheria
      curie: NCBITaxon:9347
      label: "Eutheria"
      negated: false
    notes: "INCORRECT. PIGL (N-acetylglucosaminyl-phosphatidylinositol de-N-acetylase, GlcNAc-PI -> GlcN-PI), Eutheria. Downstream of PI, consumes it. Better BP: GO:0006506 GPI anchor biosynthetic process."
  - number: 23
    conditions:
    - condition_type: FUNFAM
      value: 3.40.720.10:FF:000041
      curie: CATH.FunFam:3.40.720.10:FF:000041
      label: "GPI ethanolamine phosphate transferase 3"
      negated: false
    - condition_type: TAXON
      value: Euarchontoglires
      curie: NCBITaxon:314146
      label: "Euarchontoglires"
      negated: false
    notes: "INCORRECT. PIGO (GPI ethanolamine phosphate transferase 3), Euarchontoglires. Modifies GPI precursors. Better BP: GO:0006506."
  - number: 24
    conditions:
    - condition_type: FUNFAM
      value: 3.60.21.10:FF:000022
      curie: CATH.FunFam:3.60.21.10:FF:000022
      label: "Putative metallophosphoesterase 1"
      negated: false
    notes: "INCORRECT. PGAP5/MPPE1 metallophosphoesterase; removes ethanolamine phosphate from Man2 during GPI remodelling. Better BP: GO:0006506 GPI anchor biosynthetic process (or a GPI-remodelling term). No taxon restriction."
  - number: 25
    conditions:
    - condition_type: FUNFAM
      value: 3.90.550.10:FF:000036
      curie: CATH.FunFam:3.90.550.10:FF:000036
      label: "Dolichol-phosphate mannosyltransferase subunit 1"
      negated: false
    notes: "INCORRECT - THE SET FLAGGED BY CURATORS. FunFam 3.90.550.10:FF:000036 'Dolichol-phosphate mannosyltransferase subunit 1' (DPM1). DPM1 makes Dol-P-Man from GDP-mannose and dolichol-phosphate; it has no role in forming PI. This is the route by which S. pombe dpm1 (SPAC31G5.16c, UniProt O14466) received GO:0006661, the annotation questioned in geneontology/go-annotation issue #5835. Correct BP: GO:0180047 dolichol phosphate mannose biosynthetic process (already annotated for O14466), with GO:0006506 by pathway participation. The deep research states this is 'not supportable'."
  go_annotations:
  - go_id: GO:0006661
    go_label: phosphatidylinositol biosynthetic process
    aspect: P
  reviewed_protein_count: 0
  unreviewed_protein_count: 0
  created_date: '2021-10-20'
  modified_date: '2025-12-15'
  entries:
  - id: IPR000387
    label: "Tyrosine-specific protein phosphatases domain"
    type: INTERPRO
    appears_in_condition_sets:
    - 1
  - id: IPR010569
    label: "Myotubularin-like, phosphatase domain"
    type: INTERPRO
    appears_in_condition_sets:
    - 1
  - id: IPR002013
    label: "SAC domain"
    type: INTERPRO
    appears_in_condition_sets:
    - 2
  - id: 1.10.1070.11:FF:000002
    label: "Phosphatidylinositol 3-kinase catalytic subunit type 3"
    type: FUNFAM
    appears_in_condition_sets:
    - 3
  - id: 1.10.1070.11:FF:000006
    label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
    type: FUNFAM
    appears_in_condition_sets:
    - 4
  - id: 1.25.40.70:FF:000001
    label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
    type: FUNFAM
    appears_in_condition_sets:
    - 4
  - id: 2.60.40.150:FF:000041
    label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
    type: FUNFAM
    appears_in_condition_sets:
    - 4
  - id: 1.10.1070.11:FF:000001
    label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
    type: FUNFAM
    appears_in_condition_sets:
    - 5
  - id: 1.25.40.70:FF:000004
    label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit beta"
    type: FUNFAM
    appears_in_condition_sets:
    - 5
  - id: 2.60.40.150:FF:000046
    label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit"
    type: FUNFAM
    appears_in_condition_sets:
    - 5
  - id: 1.10.1070.11:FF:000003
    label: "Phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit beta"
    type: FUNFAM
    appears_in_condition_sets:
    - 6
  - id: 2.60.40.150:FF:000036
    label: "phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit beta"
    type: FUNFAM
    appears_in_condition_sets:
    - 6
  - id: 3.30.1010.10:FF:000001
    label: "Phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit beta"
    type: FUNFAM
    appears_in_condition_sets:
    - 6
  - id: 3.30.800.10:FF:000002
    label: "Phosphatidylinositol 5-phosphate 4-kinase type-2 beta"
    type: FUNFAM
    appears_in_condition_sets:
    - 7
  - id: 3.30.810.10:FF:000003
    label: "Phosphatidylinositol 5-phosphate 4-kinase type-2 beta"
    type: FUNFAM
    appears_in_condition_sets:
    - 7
  - id: 2.30.29.30:FF:000038
    label: "Myotubularin 1, isoform CRA_a"
    type: FUNFAM
    appears_in_condition_sets:
    - 8
  - id: 3.30.470.160:FF:000001
    label: "Kinase"
    type: FUNFAM
    appears_in_condition_sets:
    - 9
  - id: 3.30.800.10:FF:000001
    label: "phosphatidylinositol 4-phosphate 5-kinase type-1 gamma"
    type: FUNFAM
    appears_in_condition_sets:
    - 10
  - id: 3.30.800.10:FF:000009
    label: "Phosphatidylinositol 4-phosphate 5-kinase its3"
    type: FUNFAM
    appears_in_condition_sets:
    - 11
  - id: 3.40.720.10:FF:000015
    label: "GPI ethanolamine phosphate transferase 1"
    type: FUNFAM
    appears_in_condition_sets:
    - 12
  - id: 1.10.1070.11:FF:000010
    label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit gamma isoform"
    type: FUNFAM
    appears_in_condition_sets:
    - 13
  - id: 1.25.40.70:FF:000006
    label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit gamma isoform"
    type: FUNFAM
    appears_in_condition_sets:
    - 13
  - id: 2.60.40.150:FF:000087
    label: "Phosphatidylinositol 4,5-bisphosphate 3-kinase catalytic subunit gamma isoform"
    type: FUNFAM
    appears_in_condition_sets:
    - 13
  - id: 1.10.1070.11:FF:000013
    label: "Phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit gamma"
    type: FUNFAM
    appears_in_condition_sets:
    - 14
  - id: 1.25.40.70:FF:000010
    label: "Phosphatidylinositol 4-phosphate 3-kinase C2 domain-containing subunit gamma"
    type: FUNFAM
    appears_in_condition_sets:
    - 14
  - id: 1.10.287.1490:FF:000001
    label: "Putative phosphatidylinositol 3-kinase regulatory subunit alpha"
    type: FUNFAM
    appears_in_condition_sets:
    - 15
  - id: 1.10.555.10:FF:000035
    label: "Phosphatidylinositol 3-kinase regulatory subunit alpha"
    type: FUNFAM
    appears_in_condition_sets:
    - 15
  - id: 3.30.505.10:FF:000035
    label: "phosphatidylinositol 3,4,5-trisphosphate 5-phosphatase 1"
    type: FUNFAM
    appears_in_condition_sets:
    - 16
  - id: 3.60.10.10:FF:000005
    label: "phosphatidylinositol 3,4,5-trisphosphate 5-phosphatase 1"
    type: FUNFAM
    appears_in_condition_sets:
    - 16
  - id: 1.10.1070.11:FF:000019
    label: "Phosphatidylinositol 4-kinase beta 1"
    type: FUNFAM
    appears_in_condition_sets:
    - 17
  - id: 1.20.120.1760:FF:000003
    label: "CDP-diacylglycerol--inositol 3-phosphatidyltransferase"
    type: FUNFAM
    appears_in_condition_sets:
    - 18
  - id: 1.20.120.1760:FF:000021
    label: "CDP-diacylglycerol--inositol 3-phosphatidyltransferase"
    type: FUNFAM
    appears_in_condition_sets:
    - 19
  - id: 2.60.40.150:FF:000148
    label: "UV radiation resistance associated gene"
    type: FUNFAM
    appears_in_condition_sets:
    - 20
  - id: 3.30.40.10:FF:000073
    label: "myotubularin-related protein 4 isoform X2"
    type: FUNFAM
    appears_in_condition_sets:
    - 21
  - id: 3.40.50.10320:FF:000002
    label: "Probable N-acetylglucosaminyl-phosphatidylinositol de-N-acetylase"
    type: FUNFAM
    appears_in_condition_sets:
    - 22
  - id: 3.40.720.10:FF:000041
    label: "GPI ethanolamine phosphate transferase 3"
    type: FUNFAM
    appears_in_condition_sets:
    - 23
  - id: 3.60.21.10:FF:000022
    label: "Putative metallophosphoesterase 1"
    type: FUNFAM
    appears_in_condition_sets:
    - 24
  - id: 3.90.550.10:FF:000036
    label: "Dolichol-phosphate mannosyltransferase subunit 1"
    type: FUNFAM
    appears_in_condition_sets:
    - 25
  - id: GO:0006661
    label: phosphatidylinositol biosynthetic process
    type: GO_TERM
review_summary: >-
  ARBA00028655 is a wrong-branch, near-total-false-positive rule. GO:0006661 is effectively a
  leaf term denoting formation of the unphosphorylated phosphatidylinositol molecule itself
  (CDP-diacylglycerol + myo-inositol -> PI + CMP, EC 2.7.8.11). Only condition sets 18 and 19
  (CATH FunFams 1.20.120.1760:FF:000003 in Vertebrata and :FF:000021 in Fungi, both
  CDIPT/PIS) identify that enzyme. The other 23 condition sets identify PI3K/PI4K/PIP5K/PIP4K
  catalytic subunits (11 sets), phosphoinositide phosphatases (4 sets), GPI-anchor and
  dolichol-phosphate-mannose pathway enzymes (5 sets: PIGN, PIGL, PIGO, PGAP5/MPPE1, DPM1),
  and two non-catalytic regulatory subunits (PIK3R1/p85-alpha, UVRAG). Crucially, the
  phosphoinositide branch is a SIBLING of GO:0006661, not a descendant: the QuickGO ancestor
  closure of GO:0046854 "phosphatidylinositol phosphate biosynthetic process" (checked
  2026-08-15) does not contain GO:0006661 - the two terms first meet at GO:0046474
  glycerophospholipid biosynthetic process. So these are not over-broad annotations a curator
  could refine downward; they are simply in the wrong subtree. A QuickGO census on 2026-08-15
  found 4,782 GO:0006661 / ECO:0000256 / GO_REF:0000117 annotations attributed to
  ARBA00028655; in a stratified sample of 1,000 accessions only 8 (0.8%) were CDIPT/PIS,
  while 63.8% were myotubularin-family phosphoinositide 3-phosphatases and one hit was
  cytochrome b-c1 complex subunit 9 (UQCR10), a pure domain-promiscuity artifact. In the
  human subset (169 annotations, enumerated exhaustively) only 3 (1.8%) are PI synthase.
  Taxon constraints are biologically unmotivated (Homo-only INPPL1, Mus-only PIP5K1C,
  Catarrhini-only "Kinase"), and eight sets - including the DPM1 set that
  produced the S. pombe dpm1 annotation queried in geneontology/go-annotation#5835 - carry no
  taxon constraint at all. The rule is also incomplete in the one direction that would help:
  no condition set targets CDS1/CDS2, the other family legitimately part of de novo PI
  biosynthesis.
action: SPLIT
action_rationale: >-
  SPLIT rather than DEPRECATE, because condition sets 18 and 19 are correct and worth keeping:
  CDIPT/PIS is precisely the family GO:0006661 exists to describe, and a PIS-only rule would
  be a good rule. Everything else should be dissolved into the branch it actually belongs to -
  GO:0046854 (or a product-specific child such as GO:0036092) for the kinases, GO:0046856
  phosphatidylinositol dephosphorylation for the phosphatases, GO:0006506 GPI anchor
  biosynthetic process for the GPI enzymes, and GO:0180047 dolichol phosphate mannose
  biosynthetic process for DPM1 - with the non-catalytic (CS15 PIK3R1/p85, CS20 UVRAG) and
  fold-promiscuous (CS8 PH-domain superfamily, CS21 Zn-finger RING/FYVE/PHD superfamily) sets
  deleted outright rather than retargeted. If the maintainers prefer not to split, DEPRECATE
  is the correct fallback, since retaining the rule in its current form means retaining ~4,740
  wrong-branch annotations to salvage ~40 correct ones. The ~4,740 existing IEA annotations
  not derived from CS18/CS19 should be retracted. The confidence value below is confidence in
  this recommendation.
suggested_modifications:
- 'RETAIN GO:0006661 for CS18 (CATH FunFam 1.20.120.1760:FF:000003, Vertebrata) - CDIPT/PIS,
  the enzyme that forms PI. Correct as-is.'
- 'RETAIN GO:0006661 for CS19 (CATH FunFam 1.20.120.1760:FF:000021, Fungi) - PIS. Correct
  as-is. Consider merging CS18 and CS19 into a single PIS rule and relaxing the taxon to
  Eukaryota (or removing it), since the CDP-DAG/PIS route is pan-eukaryotic and PI synthase
  activity also exists outside Eukaryota.'
- 'ADD a condition set for CDS1/CDS2 (CDP-diacylglycerol synthases), currently absent: these
  supply the committed CDP-DAG precursor and are the only other family the literature
  regards as part of de novo PI biosynthesis.'
- 'RETIRE GO:0006661 from CS3, CS4, CS5, CS6, CS7, CS9, CS10, CS11, CS13, CS14 and CS17
  (PI3K class I/II/III, PI4K, PIP5K, PIP4K catalytic FunFams) and replace with GO:0046854
  phosphatidylinositol phosphate biosynthetic process; use the product-specific child where
  the FunFam pins the product, e.g. GO:0036092 phosphatidylinositol-3-phosphate biosynthetic
  process for CS3 (PIK3C3/VPS34).'
- 'RETIRE GO:0006661 from CS1 (IPR000387 + IPR010569 myotubularins), CS2 (IPR002013 SAC
  domain: SACM1L, SYNJ1/2, FIG4, INPP5F), CS16 (INPPL1/INPP5D) and CS21 (MTMR4) and replace
  with GO:0046856 phosphatidylinositol dephosphorylation. Producing PI by dephosphorylation
  is turnover, not biosynthesis.'
- 'RETIRE GO:0006661 from CS12 (PIGN), CS22 (PIGL), CS23 (PIGO) and CS24 (PGAP5/MPPE1) and
  replace with GO:0006506 GPI anchor biosynthetic process. These enzymes modify GPI
  intermediates that already contain PI.'
- 'RETIRE GO:0006661 from CS25 (CATH FunFam 3.90.550.10:FF:000036, DPM1) and replace with
  GO:0180047 dolichol phosphate mannose biosynthetic process (GO:0006506 additionally by
  pathway participation). This is the condition set that produced the S. pombe dpm1
  (SPAC31G5.16c / O14466) annotation queried in geneontology/go-annotation#5835; O14466
  already carries GO:0180047, GO:0006488, GO:0006506 and GO:0035269 and no GO:0006661.'
- 'DELETE CS15 (PIK3R1/p85-alpha regulatory subunit FunFams, Rodentia) and CS20 (UVRAG
  FunFam, Euteleostomi) entirely. Both proteins are non-catalytic; no biosynthetic BP term
  applies. At most p85 warrants the molecular function GO:0035014 phosphatidylinositol
  3-kinase regulator activity.'
- 'DELETE CS8 (CATH FunFam 2.30.29.30:FF:000038). Despite its "Myotubularin 1, isoform CRA_a"
  label this is a PH-domain-superfamily FunFam and in practice captures unrelated PH-domain
  proteins - PLEKHA3/FAPP1, PLEKHA8/FAPP2, PITPNM1 and PITPNM2 all appear in the emitted
  annotation set. A bare PH domain cannot support any catalytic BP term.'
- 'DELETE or heavily constrain CS9 (CATH FunFam 3.30.470.160:FF:000001, labelled only
  "Kinase", Catarrhini) and CS21 (CATH FunFam 3.30.40.10:FF:000073, in the promiscuous
  Zn-finger RING/FYVE/PHD superfamily, Mammalia). An uninformative label or a promiscuous
  small structural fold is not a safe sole condition.'
- 'REVIEW all taxon constraints. Homo-only (CS16), Mus-only (CS10), Catarrhini-only (CS9),
  Primates-only (CS1, CS7), Haplorrhini-only (CS2, CS8), Glires (CS14), Rodentia (CS15) and
  Eutheria/Euarchontoglires (CS22, CS23) restrictions on pan-eukaryotic enzyme families are
  annotation-bias artifacts, while CS4, CS5, CS6, CS11, CS12, CS13, CS24 and CS25 have no
  constraint at all and therefore fire across all of UniProt. The Viridiplantae restriction
  on CS17 is not in this category - plants have real PI4-kinase beta enzymes - so that set
  should be corrected on branch grounds rather than taxon grounds.'
- 'RETRACT the existing GO:0006661 IEA annotations attributed to this rule that do not derive
  from CS18/CS19 (approximately 4,740 of 4,782 as of 2026-08-15).'
parsimony:
  assessment: OVERLY_COMPLEX
  notes: >-
    25 condition sets for a single, effectively-leaf GO term is far beyond the point at which
    a rule can be reasoned about, and the complexity is not the good kind: it does not
    represent genuine biological diversity among PI-synthesising enzymes, it represents the
    accretion of every domain family that happens to co-occur with the string
    "phosphatidylinositol" in protein names. The one biologically coherent concept in the
    rule - PI synthase - is covered by just two sets (CS18, CS19), and those two are
    near-duplicates differing only in taxon (Vertebrata vs Fungi) and could be merged.
    Meanwhile the class IA/IB PI3K catalytic subunit is expressed four separate times (CS4,
    CS5, CS13, and partly CS3) and the class II PI3K twice (CS6, CS14), each as a different
    trio of FunFams from the same three CATH superfamilies (1.10.1070.11, 1.25.40.70,
    2.60.40.150). A parsimonious version of the correct part of this rule would be a single
    condition set on the CDIPT/PIS FunFam or on the corresponding InterPro PIS family entry.
  supported_by:
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-analysis.md
    supporting_text: "Summary: **2 correct (CS18, CS19)**; 15 kinase/phosphatase wrong-branch; 2 non-catalytic"
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-analysis.md
    supporting_text: "manages to be both massively over-inclusive and incomplete."
literature_support:
  assessment: CONTRADICTED
  notes: >-
    The commissioned deep research (Falcon / Edison Scientific Literature, run 2026-08-15)
    directly contradicts the rule for 23 of its 25 condition sets. Its executive conclusion is
    that GO:0006661 is not a correct blanket annotation for these families, and its per-family
    verdict table assigns "Outright incorrect" - explicitly defined there as a chemically
    different process rather than merely a less precise term - to every PI3K/PI4K/PIP5K/PIP4K
    family, every phosphoinositide phosphatase family (myotubularins, SACM1L/Sac1,
    synaptojanins, INPP5D), every GPI-pathway enzyme (PIGL, PIGN/PIGO, PGAP5/MPPE1), DPM1,
    UVRAG and PIK3R1/p85. The only families it endorses are CDIPT/PIS (the direct PI-forming
    enzyme, matching CS18 and CS19) and CDS1/CDS2 as precursor suppliers - and CDS1/CDS2 are
    not among the rule's conditions. The report is explicit that annotating DPM1 to GO:0006661
    is not supportable, which is exactly the case raised by curators in
    geneontology/go-annotation#5835. Note two honest limits of this evidence: the report found
    no retrievable GO Consortium adjudication of ARBA00028655 itself, so no formal Consortium
    ruling should be claimed; and it refers to GO:0046854 by its superseded label
    "phosphatidylinositol phosphorylation" (the current primary label, verified in QuickGO on
    2026-08-15, is "phosphatidylinositol phosphate biosynthetic process") - a labelling slip
    that does not affect its reasoning. Underlying primary and review literature (Dickson &
    Hille 2019; Blunsom & Cockcroft 2020; Fox et al. 2020; Cao et al. 2008; Kinoshita 2020;
    Imbach et al. 2000) is consistent with the report.
  supported_by:
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "GO:0006661 is not a biochemically correct blanket annotation for the listed families"
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "PIS/CDIPT is the enzyme that directly creates the PI molecule"
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "Only CDS1/CDS2 and CDIPT/PIS warrant GO:0006661 among the assessed catalytic families; CDIPT/PIS is the only direct PI-forming enzyme."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "Assigning GO:0006661 directly to these proteins conflates"
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "Kinases add phosphate to the 3-, 4-, or 5-hydroxyl of the inositol headgroup"
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "MTM1/MTMR2 use PI3P and PI(3,5)P2 as substrates, yielding PI and PI5P, respectively."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "A terminal product being PI does not convert a salvage or catabolic reaction into the canonical biosynthetic pathway."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "None produces the PI lipid moiety."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "Annotating DPM1 to GO:0006661 is not supportable."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "VPS34 is the catalytic lipid kinase and major producer of PI3P; UVRAG is a targeting/regulatory component, not a phosphoinositide-metabolizing enzyme."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "p85 has no PI/PIP lipid-kinase catalytic activity."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "established molecular function lies in a chemically different process, not merely that a more specific child term is preferable."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "Retain GO:0006661 only for condition sets that identify"
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "I found no retrievable, authoritative GO Consortium issue or publication explicitly documenting a discussion of"
condition_overlap:
  assessment: SIGNIFICANT
  notes: >-
    Two distinct overlap problems. (1) Genuine near-duplication of the same biological concept
    across sets: CS4, CS5 and CS13 each specify a trio of FunFams drawn from the same three
    CATH superfamilies (1.10.1070.11 kinase catalytic, 1.25.40.70 helical/PIK, 2.60.40.150 C2)
    for class IA/IB PI3K catalytic subunits, and CS6/CS14 do the same for class II PI3K; the
    single superfamily 1.10.1070.11 appears in seven different condition sets (CS3, CS4, CS5,
    CS6, CS13, CS14, CS17), 2.60.40.150 in five (CS4, CS5, CS6, CS13, CS20) and 3.30.800.10 in
    three (CS7 PIP4K2B, CS10 PIP5K1C, CS11 Its3). CS18/CS19 are
    the same FunFam superfamily 1.20.120.1760 split only by taxon. (2) Overlap in the emitted
    protein set even where the conditions are formally disjoint: CS1 (IPR000387 + IPR010569)
    and CS2 (IPR002013 SAC domain) both hit synaptojanins, which carry a SAC domain alongside
    a 5-phosphatase domain, and CS8's PH-domain FunFam overlaps the lipid-transfer proteins
    also reachable through other sets. Because every set emits the same single GO term, this
    redundancy does not change the annotation content - it just multiplies the number of routes
    by which a wrong annotation can be produced and makes the rule harder to repair
    incrementally. Note that no automated pairwise_overlap statistics were computed for this
    review (the `just analyze-rule` step was not run for this rule), so the overlaps described
    here are structural/biological rather than quantitative.
  supported_by:
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-analysis.md
    supporting_text: "**INCORRECT** (also ~duplicate of CS4)"
go_specificity:
  assessment: MISMATCHED
  notes: >-
    Not merely too broad - wrong branch. GO:0006661 is defined as "The chemical reactions and
    pathways resulting in the formation of phosphatidylinositol, any glycophospholipid in which
    the sn-glycerol 3-phosphate residue is esterified to the 1-hydroxyl group of 1D-myo-inositol",
    and it is effectively a leaf (its only children are the regulation terms). It therefore
    asserts specifically that the protein helps form the unphosphorylated PI molecule. The
    products of the PI kinases fall under GO:0046854 phosphatidylinositol phosphate biosynthetic
    process, whose QuickGO ancestor closure (is_a + part_of, checked 2026-08-15) is GO:0046474,
    GO:0046486, GO:0006650, GO:0006644, GO:0008654, GO:0045017, GO:0008610, GO:0006629,
    GO:0019637, GO:0006793, GO:0090407, GO:0009058, GO:0044238, GO:0008152, GO:0009987,
    GO:0008150 - GO:0006661 is absent, and the two terms first meet at GO:0046474
    glycerophospholipid biosynthetic process. Phosphoinositide kinase annotations under this
    rule are therefore in a sibling subtree, not a descendant one, and cannot be repaired by
    refining the term downward. The same argument applies to the phosphatases (whose process is
    GO:0046856 phosphatidylinositol dephosphorylation, under the dephosphorylation/catabolism
    branch), to the GPI enzymes (GO:0006506) and to DPM1 (GO:0180047). For CS18/CS19 alone the
    term is exactly right. All replacement term labels were verified against QuickGO on
    2026-08-15.
  supported_by:
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "separate phosphorylation, signaling, and GPI terms are strong evidence that GO:0006661 should not be used as a generic parent for every reaction involving a PI-containing molecule."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-analysis.md
    supporting_text: "**GO:0006661 does not appear in that list.**"
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-analysis.md
    supporting_text: "it is a **wrong-branch (MISMATCHED)** annotation whose true term lies in a parallel subtree."
taxonomic_scope:
  assessment: TOO_NARROW
  notes: >-
    The taxon constraints are internally incoherent rather than uniformly wrong, so no single
    enum value fits cleanly; TOO_NARROW is recorded because the dominant, systematic defect is
    that pan-eukaryotic enzyme families are pinned to arbitrary narrow clades. Across the 25
    sets the constraints are Primates (CS1, CS7), Haplorrhini (CS2, CS8), Catarrhini (CS9),
    Homo (CS16), Mus (CS10), Glires (CS14), Rodentia (CS15), Eutheria (CS22), Euarchontoglires
    (CS23), Mammalia (CS21), Vertebrata (CS18), Euteleostomi (CS20), Viridiplantae (CS17),
    Fungi (CS19), Eukaryota (CS3) - and none at all for CS4, CS5, CS6, CS11, CS12, CS13, CS24
    and CS25. Myotubularins, SAC-domain phosphatases, PI3Ks, PIP5Ks, the GPI pathway and DPM1
    are all pan-eukaryotic, so a Homo-only INPPL1 set or a Mus-only PIP5K1C set reflects which
    clade happened to be densely annotated in the training data, not where the biology exists.
    Conversely the eight unconstrained sets fire across all of UniProt - which is precisely how
    S. pombe dpm1 (SPAC31G5.16c) acquired GO:0006661 via CS25 and triggered
    geneontology/go-annotation#5835. The Viridiplantae restriction on CS17 is a genuine
    exception to this pattern and is not counted as an artifact here: plants do have
    PI4-kinase beta enzymes, so that constraint matches the family it names. Even the two correct
    sets are narrower than the biology: the CDP-DAG/PIS route is conserved across eukaryotes
    and PI synthase activity is not eukaryote-exclusive, so Vertebrata (CS18) and Fungi (CS19)
    could reasonably be relaxed to Eukaryota or removed.
  supported_by:
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "PI synthesis is not eukaryote-exclusive."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
    supporting_text: "its role is Dol-P-Man production for glycosylation/GPI pathways, not PI formation."
  - reference_id: file:rules/arba/ARBA00028655/ARBA00028655-analysis.md
    supporting_text: "None of these restrictions tracks the biology."
confidence: 0.95
references:
- id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
  title: Deep research analysis via Falcon (Edison Scientific Literature) - is GO:0006661 the
    correct BP term for the ARBA00028655 families?
  findings:
  - statement: Only CDIPT/PIS (and upstream CDS1/CDS2) belong in GO:0006661; the PI/PIP
      kinases, phosphoinositide phosphatases, GPI enzymes, DPM1, UVRAG and PIK3R1/p85 are
      each judged "Outright incorrect" for this term.
    supporting_text: "Only CDS1/CDS2 and CDIPT/PIS warrant GO:0006661 among the assessed catalytic families; CDIPT/PIS is the only direct PI-forming enzyme."
  - statement: The report is explicit that DPM1 should not carry GO:0006661 - the exact case
      raised by curators for S. pombe dpm1 in geneontology/go-annotation#5835.
    supporting_text: "Annotating DPM1 to GO:0006661 is not supportable."
  - statement: The recommendation for this rule is to keep GO:0006661 only for CDIPT/PIS
      condition sets and remove or replace it elsewhere.
    supporting_text: "Retain GO:0006661 only for condition sets that identify"
  - statement: No GO Consortium adjudication of ARBA00028655 itself could be retrieved, so no
      formal Consortium ruling is claimed here.
    supporting_text: "I found no retrievable, authoritative GO Consortium issue or publication explicitly documenting a discussion of"
- id: file:rules/arba/ARBA00028655/ARBA00028655-analysis.md
  title: ARBA00028655 analysis - QuickGO census of emitted annotations, per-condition-set
    verdict table, and the GO:0006661 vs GO:0046854 ontology-branch argument
  findings:
  - statement: A QuickGO census on 2026-08-15 found 4,782 GO:0006661 / ECO:0000256 /
      GO_REF:0000117 annotations attributed to ARBA00028655; in a stratified 1,000-accession
      sample only 8 (0.8%) were CDP-diacylglycerol--inositol 3-phosphatidyltransferase.
    supporting_text: "**Only 8 of 1,000 sampled proteins (0.8%) are PI synthase.**"
  - statement: In the exhaustively enumerated human subset only 3 of 169 annotations are PI
      synthase (two named CDIPT entries plus one unnamed TrEMBL PI-synthase record); every
      other human entry hit is a phosphoinositide phosphatase or a lipid-transfer/PH-domain
      protein.
    supporting_text: "**3 of 169 human annotations (1.8%) are PI synthase**"
  - statement: GO:0046854 is not a descendant of GO:0006661, so PI-kinase annotations under
      this rule are wrong-branch rather than merely imprecise.
    supporting_text: "**GO:0006661 does not appear in that list.**"
- id: PMID:30617162
  title: 'Understanding phosphoinositides: rare, dynamic, and essential membrane phospholipids.'
  findings:
  - statement: Phosphoinositides are reversibly phosphorylated derivatives of PI; kinases add
      and phosphatases remove headgroup phosphates, a process chemically distinct from
      assembling PI from CDP-DAG and myo-inositol.
- id: PMID:32117988
  title: 'CDP-Diacylglycerol Synthases (CDS): Gateway to Phosphatidylinositol and Cardiolipin
    Synthesis.'
  findings:
  - statement: De novo PI synthesis proceeds via CDS1/CDS2 (phosphatidic acid + CTP ->
      CDP-diacylglycerol) followed by PI synthase; CDP-DAG also feeds phosphatidylglycerol and
      cardiolipin synthesis.
- id: PMID:32677674
  title: 'Class IA PI3K regulatory subunits: p110-independent roles and structures.'
  findings:
  - statement: p85 regulatory subunits have no lipid-kinase catalytic activity; they stabilise
      and restrain the p110 catalytic subunit and recruit the holoenzyme to receptors -
      relevant to condition set 15.
- id: PMID:18524850
  title: Sequential actions of myotubularin lipid phosphatases regulate endosomal PI(3)P and
    growth factor receptor trafficking.
  findings:
  - statement: Myotubularins act as endosomal phosphoinositide 3-phosphatases regulating PI3P
      levels, i.e. phosphoinositide turnover rather than PI biosynthesis - relevant to
      condition sets 1, 8 and 21.
- id: PMID:32156170
  title: Biosynthesis and biology of mammalian GPI-anchored proteins.
  findings:
  - statement: GPI-anchor biosynthesis begins by modifying pre-existing PI (PIGL, PIGN/PIGO,
      PGAP5/MPPE1 all act on GPI intermediates), so these enzymes consume rather than
      synthesise PI - relevant to condition sets 12, 22, 23 and 24.
- id: PMID:10642602
  title: Deficiency of dolichol-phosphate-mannose synthase-1 causes congenital disorder of
    glycosylation type Ie.
  findings:
  - statement: DPM1 deficiency reduces dolichol-phosphate-mannose synthase activity and impairs
      N-glycosylation and GPI-anchored protein production; DPM1's reaction is GDP-mannose +
      dolichol-phosphate -> Dol-P-Man, with no role in forming PI - relevant to condition set 25.
supported_by:
- reference_id: file:rules/arba/ARBA00028655/ARBA00028655-deep-research-falcon.md
  supporting_text: "GO:0006661 is not a biochemically correct blanket annotation for the listed families"
- reference_id: file:rules/arba/ARBA00028655/ARBA00028655-analysis.md
  supporting_text: "**2 of 25 condition sets are biochemically correct; under 1% of the annotations the rule actually emits are defensible.**"