ARBA00022912

id: ARBA00022912
description: 'An extremely complex mega-rule with 210 condition sets attempting to annotate all protein phosphatases across multiple mechanistic families (PTPs, PPPs, PP2C, DUSPs) with only keyword annotation. The rule spans fundamentally different catalytic mechanisms (metal-dependent, cysteine-based, dual-specificity) and lacks essential GO term annotations for enzyme function.'
status: COMPLETE
rule_type: ARBA
rule:
  rule_id: ARBA00022912
  condition_sets: []  # 210 condition sets covering 87 InterPro domains (too complex to list here)
  go_annotations: []  # Critical issue: NO GO annotations, only keyword KW-0904
  reviewed_protein_count: 0
  unreviewed_protein_count: 353022
  created_date: '2020-05-12'
  modified_date: '2025-05-15'
review_summary: 'ARBA00022912 represents the most complex rule in the ARBA system, attempting to unify all protein phosphatases with 210 condition sets covering 87 unique InterPro domains. While the biological basis is sound (all targets are protein phosphatases), the rule suffers from extreme complexity making validation impossible, mechanistic incoherence by combining metal-dependent PPPs, cysteine-based PTPs, and dual-specificity phosphatases, and complete absence of GO term annotations despite covering enzymes with diverse molecular functions and regulatory mechanisms. The rule requires decomposition into mechanistically coherent, family-specific rules with appropriate GO annotations.'
action: MODIFY
action_rationale: 'While the core biological concept is correct (all targets are protein phosphatases), the rule design is fundamentally flawed. The 210 condition sets create unmanageable complexity and attempt to unify mechanistically distinct phosphatase families that differ in catalytic mechanism, cofactor requirements, and substrate specificity. Most critically, the rule provides only keyword annotation (KW-0904) without essential GO terms like GO:0004721 (phosphoprotein phosphatase activity), GO:0004725 (protein tyrosine phosphatase activity), and GO:0006470 (protein dephosphorylation). The rule must be decomposed into focused, family-specific rules with appropriate molecular function and biological process annotations.'
suggested_modifications:
- 'Decompose into mechanism-based rules: separate rules for PTP family (tyrosine phosphatases), PPP family (serine/threonine phosphatases), PP2C family, and dual-specificity phosphatases'
- 'Add essential GO annotations: GO:0016791 (phosphatase activity), GO:0006470 (protein dephosphorylation), plus family-specific molecular function terms'
- 'Reduce condition sets per rule to <20 for maintainability and validation'
- 'Implement family-specific taxonomic boundaries based on phylogenetic distribution'
- 'Add negative conditions to exclude known pseudophosphatases'
- 'Consider substrate specificity differences when designing replacement rules'
parsimony:
  assessment: OVERLY_COMPLEX
  notes: 'With 210 condition sets covering 87 unique InterPro domains across all phosphatase families, this rule far exceeds any reasonable complexity threshold. The rule attempts to unify mechanistically distinct phosphatase families (PTPs, PPPs, PP2C, DUSPs) that differ fundamentally in catalytic mechanism, cofactor requirements, and substrate specificity. This massive scope violates the principle that annotation rules should target coherent, mechanistically related protein sets.'
  supported_by:
  - reference_id: file:rules/arba/ARBA00022912/ARBA00022912-deep-research-manual.md
    supporting_text: 'ARBA00022912 is an extremely complex mega-rule attempting to annotate all protein phosphatases with 210 condition sets covering 87 unique InterPro domains. The rule spans multiple mechanistically distinct phosphatase families and provides only keyword annotation (KW-0904 "Protein phosphatase") without essential GO term annotations.'
literature_support:
  assessment: MODERATE
  notes: 'Protein phosphatases are extensively characterized enzymes with strong literature support for their catalytic activities. However, the literature overwhelmingly supports family-specific classification based on distinct catalytic mechanisms, cofactor requirements, and substrate specificities. Key reviews demonstrate that phosphatase families arose through convergent evolution and require separate annotation approaches rather than this unified mega-rule strategy.'
  supported_by:
  - reference_id: file:rules/arba/ARBA00022912/ARBA00022912-deep-research-manual.md
    supporting_text: 'The protein phosphatase field has well-established classification systems: (1) Alonso et al. (2004) "Protein tyrosine phosphatases in the human genome" - Definitive classification of 107 human PTPs, shows clear mechanistic families requiring separate annotation (2) Shi (2009) "Serine/threonine phosphatases: mechanism through structure" - Establishes mechanistic basis for PPP family classification'
condition_overlap:
  assessment: SIGNIFICANT
  notes: 'With 210 condition sets and 87 unique InterPro domains spanning all phosphatase families, extensive overlap is inevitable. Many phosphatase domains appear across multiple families, and some domains are found in non-phosphatase proteins. The rule combines conditions for mechanistically distinct families with different catalytic mechanisms, creating complex overlap patterns that increase false positive risk.'
  supported_by:
  - reference_id: file:rules/arba/ARBA00022912/ARBA00022912-deep-research-manual.md
    supporting_text: 'The rule combines domains from mechanistically distinct families: Metal-dependent PP2A family (requires Mn2+ or Mg2+), Cysteine-based PTPs (acid/base catalysis, no metal requirement), Ca2+/calmodulin-dependent PP2B (calcineurin), PP2C family (Mg2+-dependent, different fold)'
go_specificity:
  assessment: MISSING
  notes: 'Critical issue: The rule provides NO GO term annotations, only keyword KW-0904 "Protein phosphatase". This is completely inadequate for enzyme annotation. Protein phosphatases should minimally have GO:0016791 (phosphatase activity) and GO:0006470 (protein dephosphorylation). Family-specific rules should include more specific molecular function terms reflecting catalytic mechanism and substrate specificity.'
  supported_by:
  - reference_id: file:rules/arba/ARBA00022912/ARBA00022912-deep-research-manual.md
    supporting_text: 'Critical issue: The rule provides NO GO term annotations, only keyword KW-0904 "Protein phosphatase". Minimum required: GO:0016791 (phosphatase activity), GO:0006470 (protein dephosphorylation). Family-specific: GO:0004725 (protein tyrosine phosphatase activity) for PTPs, GO:0004721 (phosphoprotein phosphatase activity) for Ser/Thr phosphatases'
taxonomic_scope:
  assessment: TOO_BROAD
  notes: 'The rule applies universally to all taxa without considering family-specific phylogenetic distributions. Some phosphatase families are taxon-specific, and prokaryotic vs eukaryotic phosphatases often have different regulatory mechanisms and cellular functions. Universal application risks inappropriate cross-kingdom annotations and fails to capture organism-specific phosphatase biology.'
  supported_by:
  - reference_id: file:rules/arba/ARBA00022912/ARBA00022912-deep-research-manual.md
    supporting_text: 'Current Scope: Universal (all taxa). Problems: Some phosphatase families are taxon-specific, Prokaryotic vs eukaryotic phosphatases have different functions, Plant-specific phosphatases need distinct annotation. Recommended: Family-specific taxonomic boundaries, Consider organism-specific regulatory mechanisms'
confidence: 0.9
additional_notes: 'This rule represents an extreme example of over-ambitious scope in automated annotation. While the goal of systematic phosphatase annotation is laudable, the execution creates more problems than it solves. The 210 condition sets make human validation impossible, and the lack of GO annotations severely limits the utility of any annotations produced. Given the well-established literature on phosphatase classification, this rule should be completely redesigned as a set of mechanistically coherent, family-specific rules with appropriate GO annotations. The false positive risk from such a broad rule is substantial, particularly for pseudophosphatases and multi-domain proteins with phosphatase-like domains but different primary functions.'
entries: []  # Would be populated with pairwise overlap analysis if analysis tools were available