id: ARBA00022759 description: An extremely complex rule with 318 condition sets that all predict the same broad "Endonuclease" keyword annotation (KW-0255). This rule attempts to capture all types of nucleic acid cleaving enzymes through diverse combinations of InterPro domains, PANTHER families, and taxonomic restrictions, representing a clear example of rule over-engineering that conflicts with annotation best practices. status: COMPLETE rule_type: ARBA rule: rule_id: ARBA00022759 condition_sets: - number: 1 conditions: - condition_type: INTERPRO value: IPR001352 curie: InterPro:IPR001352 label: Kyprides_PLpro-like negated: false - condition_type: INTERPRO value: IPR012337 curie: InterPro:IPR012337 label: Ribonuclease H negated: false - condition_type: INTERPRO value: IPR024567 curie: InterPro:IPR024567 label: Ribonuclease HII/HIII domain negated: false notes: Representative example - targets RNase H enzymes through multiple domain requirements. While individually reasonable domains, the AND logic may be overly restrictive and this represents just 1 of 318 similar condition sets. - number: 2 conditions: - condition_type: INTERPRO value: IPR002036 curie: InterPro:IPR002036 label: SRP54-type protein, GTPase domain negated: false - condition_type: INTERPRO value: IPR023091 curie: InterPro:IPR023091 label: Ribosomal protein S30AE signature negated: false notes: Highly problematic combination - these domains are associated with ribosomal proteins and signal recognition particles, not endonucleases. Clear example of false positive risk from overly broad domain inclusion. - number: "3-318" conditions: [] notes: 316 additional condition sets with similar patterns - combinations of InterPro domains, PANTHER families, and taxonomic restrictions that attempt to capture the enormous diversity of endonuclease enzymes. The sheer number makes individual review impractical and creates an unmaintainable annotation rule. go_annotations: [] keyword_annotations: - kw_id: KW-0255 kw_label: Endonuclease category: Molecular function reviewed_protein_count: 0 unreviewed_protein_count: 0 created_date: '2020-05-12' modified_date: '2025-05-15' review_summary: This rule represents a catastrophic failure of annotation rule design. With 318 condition sets all predicting the same overly broad "Endonuclease" keyword, it violates every principle of parsimonious rule construction. The functional category is so broad as to be biologically meaningless, encompassing enzymes with completely different mechanisms, substrates, and cellular roles. Many condition sets include domains from non-nuclease proteins, creating massive false positive risk. The rule is impossible for humans to review, maintain, or validate. This should serve as a cautionary example of how not to design annotation rules. action: REMOVE action_rationale: This rule must be REMOVED entirely because it represents a fundamentally flawed approach to functional annotation. The 318 condition sets make it impossible to review or maintain, while the overly broad "Endonuclease" classification provides no meaningful biological insight. Many condition sets include promiscuous domains that will generate false positives, and the lack of mechanistic or functional specificity conflicts with modern annotation best practices. The rule should be replaced with 10-15 subfamily-specific rules that provide precise functional classifications for distinct nuclease families (RNase H, restriction endonucleases, CRISPR nucleases, etc.) using appropriate GO molecular function terms instead of generic keywords. suggested_modifications: - Replace with subfamily-specific rules for major endonuclease families - Use GO molecular function terms (GO:0004519 and specific child terms) instead of broad keywords - Limit condition sets to <10 per rule for maintainability - Include mechanistic specificity (metal-dependent vs independent, RNA vs DNA specificity) - Add pathway context (DNA repair, RNA processing, defense, etc.) - Implement systematic review process for domain promiscuity parsimony: assessment: OVERLY_COMPLEX notes: This rule violates every principle of parsimony with 318 condition sets that could be reduced to 10-15 subfamily-specific rules. The massive complexity serves no biological purpose and creates an unmaintainable annotation system. The rule represents the antithesis of parsimonious design, attempting to solve a complex classification problem through brute-force enumeration rather than systematic functional analysis. supported_by: - reference_id: file:rules/arba/ARBA00022759/ARBA00022759-analysis-notes.md supporting_text: The rule contains 318 condition sets making it impossible for humans to review, maintain, or validate. This represents an extreme outlier compared to typical ARBA rules with 1-20 condition sets. literature_support: assessment: CONTRADICTED notes: Literature strongly contradicts the approach of grouping all endonucleases under a single broad annotation. Functional genomics studies emphasize the critical importance of subfamily-level classification for nucleases, as they have fundamentally different mechanisms, substrates, and biological roles. The scientific consensus supports mechanistic and pathway-specific classification rather than broad functional grouping. supported_by: - reference_id: file:rules/arba/ARBA00022759/ARBA00022759-deep-research-manual.md supporting_text: Bacterial ribonucleases alone include >20 distinct families with different substrate specificities, cellular roles, and essentiality patterns. Grouping all under "endonuclease" obscures critical functional distinctions. - reference_id: file:rules/arba/ARBA00022759/ARBA00022759-deep-research-manual.md supporting_text: DNA repair endonucleases operate in distinct pathways - base excision repair (APE1, APE2), nucleotide excision repair (UvrABC), mismatch repair (MutH), double-strand break repair (Mre11, CtIP). Each has specific substrate requirements and pathway integration. condition_overlap: assessment: SIGNIFICANT notes: The 318 condition sets show massive redundancy and overlap, with many targeting the same or highly similar protein sets through different domain combinations. This creates computational waste and annotation conflicts. Many domains appear in multiple condition sets, suggesting systematic redundancy throughout the rule structure. supported_by: - reference_id: file:rules/arba/ARBA00022759/ARBA00022759-analysis-notes.md supporting_text: Several PANTHER subfamilies overlap significantly in protein coverage, suggesting redundant condition sets within the rule. go_specificity: assessment: TOO_BROAD notes: The rule uses a keyword annotation rather than specific GO terms, which conflicts with modern annotation best practices. "Endonuclease" is far too broad - it should use GO:0004519 (endonuclease activity) and specific child terms for each nuclease subfamily (e.g., GO:0004523 for ribonuclease H activity, GO:0015643 for restriction enzyme activity). The lack of mechanistic or pathway specificity makes the annotation nearly useless for biological interpretation. supported_by: - reference_id: file:rules/arba/ARBA00022759/ARBA00022759-deep-research-manual.md supporting_text: Better served by GO molecular function terms instead of keywords - GO:0004519 (endonuclease activity), GO:0003676 (nucleic acid binding), and specific child terms for each nuclease type. taxonomic_scope: assessment: INAPPROPRIATE notes: The rule includes taxonomic restrictions in multiple condition sets but applies them inconsistently and often inappropriately. Some condition sets restrict to specific viral taxa for unclear reasons, while others have no taxonomic scope despite targeting lineage-specific enzyme families. The taxonomic logic appears ad hoc rather than based on systematic evolutionary analysis. supported_by: - reference_id: file:rules/arba/ARBA00022759/ARBA00022759-analysis-notes.md supporting_text: Condition Set 10 includes viral restriction (IPR043128 + taxon Pararnavirae) with unclear functional justification, and viral proteins may have nuclease activity as secondary function. confidence: 0.05 references: - id: file:rules/arba/ARBA00022759/ARBA00022759-deep-research-manual.md title: Deep research analysis on endonuclease functional diversity findings: - statement: Endonucleases represent an extremely diverse functional class with multiple evolutionary origins, catalytic mechanisms, and biological roles that should not be grouped under a single annotation. - statement: The rule mixes proteins with fundamentally different mechanisms - metal-dependent vs metal-independent, single-strand vs double-strand cleavage, RNA vs DNA specificity, sequence-specific vs non-specific. - statement: Many InterPro domains in this rule are non-specific, structural, or promiscuous, found in proteins with other functions and creating false positive risk. - id: file:rules/arba/ARBA00022759/ARBA00022759-analysis-notes.md title: Quantitative analysis of rule complexity and functional problems findings: - statement: The rule contains 318 condition sets with ~150+ unique InterPro domains and ~30+ PANTHER families, making it an extreme outlier in complexity compared to normal ARBA rules with 1-20 condition sets. - statement: Several domains appear in condition sets but represent non-nuclease functions - SAM-dependent methyltransferases, histidine kinases, xylose isomerases. - statement: The rule shows massive redundancy with PANTHER subfamilies overlapping significantly in protein coverage and many InterPro domains having existing InterPro2GO mappings. - id: UniProt:KW-0255 title: Endonuclease keyword definition findings: - statement: UniProt keyword for enzymes that cleave nucleic acid chains at internal positions, representing an overly broad functional classification unsuitable for precise annotation. supported_by: - reference_id: file:rules/arba/ARBA00022759/ARBA00022759-deep-research-manual.md supporting_text: This rule should be REMOVED entirely because excessive complexity makes maintenance impossible, functional category is too broad to be informative, high risk of false positives from promiscuous domains, conflicts with GO annotation best practices requiring specific functional terms.