date: '2026-09-20'
reviewer: codex
genes:
- gene_file: genes/PSEPK/PP_3394/PP_3394-ai-review.yaml
  status: awaiting_adjudication
  scope: full_gene
  annotation_count: 6
  outcome: Restored broad catalysis; specific cleavage/synthase and pathway assignments remain unresolved pending
    reaction adjudication.
  changes:
  - index: 0
    term: GO:0003824
    label: catalytic activity
    evidence_type: IEA
    reference: GO_REF:0000002
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: ACCEPT
    reason: The DRE-TIM/HMGL-like and alpha-isopropylmalate/homocitrate-synthase signatures imply catalysis. Their
      disagreement concerns substrate and reaction class, not whether PP_3394 is catalytic; no target-specific loss
      of catalysis has been established.
  - index: 1
    term: GO:0004419
    label: hydroxymethylglutaryl-CoA lyase activity
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: UNDECIDED
    new_action: UNDECIDED
    reason: An alpha-isopropylmalate/homocitrate-synthase signature creates a paralog-specificity conflict that
      a first pass cannot resolve. The existing BCAA module report explicitly leaves PP_3394 versus mvaB unresolved;
      the mvaB report also does not assign a reaction to PP_3394. TreeGrafter source PTN000031316 is recorded as
      an unresolved transfer, not a demonstrated target-specific loss.
  - index: 2
    term: GO:0006552
    label: L-leucine catabolic process
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: UNDECIDED
    new_action: UNDECIDED
    reason: Do not infer a leucine-catabolic role from the shared fold alone. The existing BCAA module report explicitly
      leaves PP_3394 versus mvaB unresolved; the mvaB report also does not assign a reaction to PP_3394. TreeGrafter
      source PTN000031316 is recorded as an unresolved transfer, not a demonstrated target-specific loss.
  - index: 3
    term: GO:0016833
    label: oxo-acid-lyase activity
    evidence_type: IEA
    reference: GO_REF:0000120
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: UNDECIDED
    reason: GO:0016833 specifies nonhydrolytic, nonoxidative C-C cleavage of a 3-hydroxy acid (QuickGO definition
      checked 2026-09-20). HMG-CoA cleavage fits, whereas a synthase-family assignment alone does not establish
      this cleavage activity. This uncertainty is about chemistry, not excessive generality.
  - index: 5
    term: GO:0046951
    label: ketone body biosynthetic process
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: UNDECIDED
    new_action: UNDECIDED
    reason: The underlying molecular-function assignment is unresolved. The existing BCAA module report explicitly
      leaves PP_3394 versus mvaB unresolved; the mvaB report also does not assign a reaction to PP_3394. TreeGrafter
      source PTN000031316 is recorded as an unresolved transfer, not a demonstrated target-specific loss.
  evidence:
    summary: 'Catalytic activity is retained: uncertainty between HMGL-like and synthase-family reaction assignments
      does not establish loss of catalysis. Oxo-acid lyase is separately UNDECIDED because its definition specifies
      C-C cleavage of a 3-hydroxy acid, not any DRE-TIM reaction. The existing mvaB report and the actual BCAA module
      report both leave PP_3394 unresolved. The latter explicitly states: "Both are candidates for the displaced
      leucine terminal step; choice is **unresolved** and should be flagged." TreeGrafter PTN000031316 supports
      the original HMG-CoA-lyase/leucine/ketone annotations, but the target-specific reaction is not settled by
      a paralog comparison. A focused reaction adjudication is coordinated by the parent reviewer; no duplicate
      was launched.'
    references:
    - file:PSEPK/PP_3394/PP_3394-uniprot.txt
    - PMID:19459965
    - file:PSEPK/mvaB/mvaB-deep-research-openscientist.md
    - file:projects/TREEGRAFTER/rereview-2026-09-20/cache-reports/openscientist-commissioned-modulepathwaytaxo-d2b7396e.md
    cache_provenance: projects/TREEGRAFTER/rereview-2026-09-20/cache-reports/export-provenance.json
  adjudication:
    status: focused reaction report requested by root
    duplicate_request_launched: false
  validation:
    command: just validate PSEPK PP_3394
    status: passed
    source_metadata_preserved: true
- gene_file: genes/PSEPK/alg8/alg8-ai-review.yaml
  status: reviewed
  scope: full_gene
  annotation_count: 6
  outcome: Restored conditional matrix assembly and corrected substrate/strain-transfer reasoning.
  changes:
  - index: 1
    term: GO:0030213
    label: hyaluronan biosynthetic process
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: REMOVE
    new_action: REMOVE
    reason: Alg8 polymerizes GDP-mannuronate into the alginate precursor mannuronan. Hyaluronan instead contains
      alternating glucuronic acid and N-acetylglucosamine. The rejection rests on substrate specificity, not a claim
      that hyaluronan synthesis is restricted to vertebrates.
  - index: 3
    term: GO:0085029
    label: extracellular matrix assembly
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: KEEP_AS_NON_CORE
    reason: GO:0085029 covers aggregation, arrangement and bonding of extracellular matrix. Alg8 performs polymer
      assembly rather than merely being a substrate of the process. PMID:17601783 links alginate to biofilm architecture
      under water limitation in P. putida mt-2; this supports conditional matrix function by related-strain evidence.
      PMID:25968647 establishes the Alg8-Alg44 polymerase mechanism in P. aeruginosa. Transfer to the conserved
      KT2440 machinery is reasonable, while alginic acid biosynthesis remains the most specific core process.
  evidence:
    summary: Extracellular matrix assembly is retained as a conditional non-core context of alginate polymerization.
      Alg8 performs polymer-building work, and a broader process need not be its most specific core description.
      PMID:17601783 states "Alginate influenced biofilm architecture" but explicitly assays P. putida mt-2, not
      KT2440. PMID:25968647 directly assays the P. aeruginosa Alg8-Alg44 complex. The existing OpenScientist operon
      synthesis was read and incorporated with this strain-transfer caveat; its direct-KT2440 attribution is disputed.
      Hyaluronan terms remain incorrect because Alg8 uses GDP-mannuronate for alginate rather than the alternating
      hyaluronan substrates; the erroneous vertebrate-only rationale was removed. No source annotation was deleted
      and no new term was added.
    references:
    - file:PSEPK/alg8/alg8-uniprot.txt
    - file:PSEPK/alg8/alg8-goa.tsv
    - file:interpro/panther/PTHR22913/PTHR22913-entries.csv
    - PMID:17601783
    - PMID:25968647
    - file:projects/P_PUTIDA/deep-research/PSEPK__alginate-polymerization-export__ppu00543-deep-research-openscientist.md
    - file:modules/alginate_polymerization_export.yaml
    cache_provenance: null
  adjudication:
    status: existing evidence incorporated; residual uncertainty stated in annotations
    duplicate_request_launched: false
  validation:
    command: just validate PSEPK alg8
    status: passed
    source_metadata_preserved: true
- gene_file: genes/PSEPK/purM/purM-ai-review.yaml
  status: reviewed
  scope: full_gene
  annotation_count: 7
  outcome: Restored soluble localization and valid broad purine/adenine biosynthesis; retained the specific wrong-domain
    GARS rejection.
  changes:
  - index: 0
    term: GO:0004637
    label: phosphoribosylamine-glycine ligase activity
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: REMOVE
    new_action: REMOVE
    reason: Q88MA9 contains AIRS and AIRS_C domains (Pfam PF00586/PF02769) with PurM-specific IPR004733. It is the
      standalone FGAM cyclase, whereas phosphoribosylamine-glycine ligation requires the distinct GARS/PurD domain
      present in bifunctional ADE5,7 or trifunctional GART. PTN002237315 therefore conflates different catalytic
      portions of a fused family for this molecular function; the shared pathway annotations do not inherit that
      defect.
  - index: 2
    term: GO:0005737
    label: cytoplasm
    evidence_type: IEA
    reference: GO_REF:0000120
    old_action: KEEP_AS_NON_CORE
    new_action: ACCEPT
    reason: The target UniProt record assigns cytoplasmic localization through HAMAP MF_00741, and the standalone
      soluble AIR-synthetase architecture and conserved intracellular purine pathway support the inference. A direct
      localization assay is not required to retain a sound family-based location, and breadth does not make cytoplasm
      non-core.
  - index: 3
    term: GO:0005829
    label: cytosol
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: KEEP_AS_NON_CORE
    new_action: ACCEPT
    reason: The target UniProt record assigns cytoplasmic localization through HAMAP MF_00741, and the standalone
      soluble AIR-synthetase architecture and conserved intracellular purine pathway support the inference. A direct
      localization assay is not required to retain a sound family-based location, and breadth does not make cytoplasm
      non-core.
  - index: 4
    term: GO:0006164
    label: purine nucleotide biosynthetic process
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: MODIFY
    new_action: ACCEPT
    reason: FGAM-to-AIR cyclization is part of the route to IMP, a purine nucleotide. The broader process is correct
      and compatible with the more specific de novo IMP annotation; a duplicate narrower replacement is unnecessary.
  - index: 6
    term: GO:0046084
    label: adenine biosynthetic process
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: ACCEPT
    reason: 'The term is broader than a terminal free-base-release enzyme: current experimental GO:0046084 annotations
      include Candida ADE5:7 (A0A1D8PE67; PMID:17600079), an AIR-synthetase/GARS fusion, as well as yeast ADE8 and
      fission-yeast ade7. The Candida primary abstract reports an adenine-deficient deletion mutant; UniProt independently
      confirms its AIRS activity. PurM performs that conserved ring-building step. The comparator convention and
      adenine-salvage child term argue against restricting GO:0046084 to free-base release. A local IMP-module boundary
      does not negate biological participation.'
  evidence:
    summary: 'The standalone AIRS/AIRS_C architecture supports the exact PurM reaction and rejects transfer of the
      distinct PurD/GARS activity from the fused GART-family context (PTN002237315). Cytoplasm/cytosol and broad
      purine-nucleotide biosynthesis are core compatible inferences, not weak annotations because they are electronic
      or broad. GO:0046084 adenine biosynthesis was checked against current GO usage: exact experimental annotations
      include Candida ADE5:7 (A0A1D8PE67), yeast ADE8 and fission-yeast ade7. The official UniProt comparator record
      confirms AIRS activity and RHEA:23032 in ADE5:7; PMID:17600079 reports "resulting mutants were adenine deficient".
      These upstream de novo-enzyme comparators contradict a categorical restriction to free-base-release enzymes.
      PurM contributes the same AIR-synthesis work, so the adenine annotation is accepted; the project module boundary
      is not a biological exclusion. All five pre-existing cached purine reports were read: three inventory/context
      hits and two pathway-mechanism reports, none a focused GO:0046084 adjudication. Their generated contents were
      exported with ResearchProcessor and hashes, not rewritten. The live comparator evidence settles the scope
      without another OpenScientist request.'
    references:
    - file:PSEPK/purM/purM-uniprot.txt
    - file:projects/TREEGRAFTER/rereview-2026-09-20/metabolic-comparators.json
    - file:PSEPK/purM/purM-comparator-ADE5-7.json
    - PMID:17600079
    - file:projects/TREEGRAFTER/rereview-2026-09-20/cache-reports/openscientist-commissioned-review-brief-revi-1a1f035d.md
    cache_provenance: projects/TREEGRAFTER/rereview-2026-09-20/cache-reports/export-provenance.json
  adjudication:
    status: existing evidence incorporated; residual uncertainty stated in annotations
    duplicate_request_launched: false
  validation:
    command: just validate PSEPK purM
    status: passed
    source_metadata_preserved: true
- gene_file: genes/PSEPK/ilvA-I/ilvA-I-ai-review.yaml
  status: reviewed
  scope: full_gene
  annotation_count: 7
  outcome: Restored threonine breakdown and core parent/cofactor annotations; retained secondary serine capacity
    with unresolved physiological flux.
  changes:
  - index: 0
    term: GO:0003941
    label: L-serine ammonia-lyase activity
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: KEEP_AS_NON_CORE
    reason: Direct recombinant enzyme studies demonstrate L-serine use by biosynthetic IlvA homologs (PMID:30226377,
      Mycobacterium smegmatis; PMID:30327426, Salmonella enterica). The target retains the biosynthetic Ser/Thr-dehydratase
      catalytic architecture, supporting the TreeGrafter molecular-activity inference even though its main pathway
      supplies isoleucine. This is comparative evidence, not a target-specific kinetic measurement or proof of physiological
      serine flux.
  - index: 2
    term: GO:0006520
    label: amino acid metabolic process
    evidence_type: IEA
    reference: GO_REF:0000002
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: ACCEPT
    reason: Its catalysis consumes L-threonine and supplies 2-oxobutanoate for L-isoleucine synthesis. Both the
      broad amino-acid process and the specific isoleucine biosynthetic process are correct; annotation breadth
      does not imply excess biological scope.
  - index: 3
    term: GO:0006565
    label: L-serine catabolic process
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: UNDECIDED
    reason: Family enzyme assays support a serine-dehydratase side activity, and PMID:30327426 shows that this chemistry
      can occur in vivo in a biosynthetic IlvA context. They do not determine the contribution of either KT2440
      paralog to physiological serine turnover. The isoleucine role does not exclude such a contribution; quantitative
      target-specific flux evidence is missing.
  - index: 4
    term: GO:0006567
    label: L-threonine catabolic process
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: ACCEPT
    reason: GO:0006567 covers chemical reactions resulting in L-threonine breakdown. Conversion to 2-oxobutanoate
      and ammonium satisfies that definition. Subsequent use of the carbon skeleton in biosynthesis does not exclude
      threonine catabolism or require a dedicated energy-generating degradation pathway.
  - index: 5
    term: GO:0030170
    label: pyridoxal phosphate binding
    evidence_type: IEA
    reference: GO_REF:0000002
    old_action: KEEP_AS_NON_CORE
    new_action: ACCEPT
    reason: The conserved pyridoxal-phosphate-binding catalytic domain and UniProt cofactor assignment support binding
      the required cofactor. The existence of a more informative reaction term does not make this essential catalytic
      property non-core.
  - index: 6
    term: GO:1901705
    label: L-isoleucine biosynthetic process
    evidence_type: IEA
    reference: GO_REF:0000002
    old_action: ACCEPT
    new_action: ACCEPT
    reason: UniProt assigns Q88HB4 to the single conversion of L-threonine to 2-oxobutanoate in L-isoleucine biosynthesis.
  evidence:
    summary: 'All seven annotations were re-reviewed with the existing OpenScientist report and primary comparative
      biochemistry. L-threonine catabolism is ACCEPT: the enzyme breaks down threonine even though its carbon skeleton
      feeds isoleucine biosynthesis. Broad amino-acid metabolism and required PLP binding are also core. Secondary
      L-serine ammonia-lyase activity is retained as inferred non-core capacity, supported by recombinant M. smegmatis
      TD (PMID:30226377) and Salmonella IlvA assays (PMID:30327426), without pretending these are KT2440 measurements.
      Physiological serine catabolism is UNDECIDED for each paralog because actual target flux remains unknown.
      The report explicitly says KT2440 serine-side-reaction physiology is untested. Its 75% identity does not date
      the duplication, and the exact E. coli 23-angstrom allosteric relay was not measured in either target. The
      net reaction follows RHEA:22108 (threonine to 2-oxobutanoate and ammonium), without the report''s extraneous
      water product. Serine generates 2-aminoacrylate whereas threonine generates 2-aminocrotonate; the primary
      paper explicitly distinguishes them.'
    references:
    - file:PSEPK/ilvA-I/ilvA-I-uniprot.txt
    - file:PSEPK/ilvA-I/ilvA-I-goa.tsv
    - file:PSEPK/ilvA-I/ilvA-I-deep-research-openscientist.md
    - PMID:30226377
    - PMID:30327426
    cache_provenance: null
  adjudication:
    status: existing evidence incorporated; residual uncertainty stated in annotations
    duplicate_request_launched: false
  validation:
    command: just validate PSEPK ilvA-I
    status: passed
    source_metadata_preserved: true
- gene_file: genes/PSEPK/ilvA-II/ilvA-II-ai-review.yaml
  status: reviewed
  scope: full_gene
  annotation_count: 7
  outcome: Restored threonine breakdown and core parent/cofactor annotations using existing comparative report;
    physiological serine flux remains unresolved.
  changes:
  - index: 0
    term: GO:0003941
    label: L-serine ammonia-lyase activity
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: KEEP_AS_NON_CORE
    reason: Direct recombinant enzyme studies demonstrate L-serine use by biosynthetic IlvA homologs (PMID:30226377,
      Mycobacterium smegmatis; PMID:30327426, Salmonella enterica). The target retains the biosynthetic Ser/Thr-dehydratase
      catalytic architecture, supporting the TreeGrafter molecular-activity inference even though its main pathway
      supplies isoleucine. This is comparative evidence, not a target-specific kinetic measurement or proof of physiological
      serine flux.
  - index: 2
    term: GO:0006520
    label: amino acid metabolic process
    evidence_type: IEA
    reference: GO_REF:0000002
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: ACCEPT
    reason: Its catalysis consumes L-threonine and supplies 2-oxobutanoate for L-isoleucine synthesis. Both the
      broad amino-acid process and the specific isoleucine biosynthetic process are correct; annotation breadth
      does not imply excess biological scope.
  - index: 3
    term: GO:0006565
    label: L-serine catabolic process
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: UNDECIDED
    reason: Family enzyme assays support a serine-dehydratase side activity, and PMID:30327426 shows that this chemistry
      can occur in vivo in a biosynthetic IlvA context. They do not determine the contribution of either KT2440
      paralog to physiological serine turnover. The isoleucine role does not exclude such a contribution; quantitative
      target-specific flux evidence is missing.
  - index: 4
    term: GO:0006567
    label: L-threonine catabolic process
    evidence_type: IEA
    reference: GO_REF:0000118
    old_action: MARK_AS_OVER_ANNOTATED
    new_action: ACCEPT
    reason: GO:0006567 covers chemical reactions resulting in L-threonine breakdown. Conversion to 2-oxobutanoate
      and ammonium satisfies that definition. Subsequent use of the carbon skeleton in biosynthesis does not exclude
      threonine catabolism or require a dedicated energy-generating degradation pathway.
  - index: 5
    term: GO:0030170
    label: pyridoxal phosphate binding
    evidence_type: IEA
    reference: GO_REF:0000002
    old_action: KEEP_AS_NON_CORE
    new_action: ACCEPT
    reason: The conserved pyridoxal-phosphate-binding catalytic domain and UniProt cofactor assignment support binding
      the required cofactor. The existence of a more informative reaction term does not make this essential catalytic
      property non-core.
  - index: 6
    term: GO:1901705
    label: L-isoleucine biosynthetic process
    evidence_type: IEA
    reference: GO_REF:0000002
    old_action: ACCEPT
    new_action: ACCEPT
    reason: UniProt assigns Q88CN1 to the single conversion of L-threonine to 2-oxobutanoate in L-isoleucine biosynthesis.
  evidence:
    summary: All seven annotations were re-reviewed using the existing IlvA-I OpenScientist report, which substantively
      compares both KT2440 paralogs; no duplicate report was requested. L-threonine catabolism, amino-acid metabolism
      and PLP binding are ACCEPT for the conserved core reaction. Secondary serine molecular activity is retained
      as comparative non-core capacity, supported by PMID:30226377 and PMID:30327426; physiological serine catabolism
      remains UNDECIDED. The two paralogs share full catalytic/ACT architecture, but their relative flux, kinetic
      and regulatory contributions are not measured. Report caveats include an undated duplication, an unmeasured
      target-specific 23-angstrom allosteric relay and an incorrect net water product; 2-aminoacrylate is serine-derived,
      whereas the threonine intermediate is 2-aminocrotonate.
    references:
    - file:PSEPK/ilvA-II/ilvA-II-uniprot.txt
    - file:PSEPK/ilvA-II/ilvA-II-goa.tsv
    - PMID:30226377
    - PMID:30327426
    - file:PSEPK/ilvA-I/ilvA-I-deep-research-openscientist.md
    cache_provenance: null
  adjudication:
    status: existing evidence incorporated; residual uncertainty stated in annotations
    duplicate_request_launched: false
  validation:
    command: just validate PSEPK ilvA-II
    status: passed
    source_metadata_preserved: true