id: ARBA00022801 description: 'ARBA rule predicting hydrolase activity based on 4070 condition sets - exceptionally complex rule requiring removal' status: COMPLETE rule_type: ARBA rule: rule_id: ARBA00022801 condition_sets: - number: 1 conditions: - condition_type: InterPro id value: IPR014014 curie: InterPro:IPR014014 label: "Beta-lactamase class A" negated: false - condition_type: InterPro id value: IPR027417 curie: InterPro:IPR027417 label: "P-loop containing nucleoside triphosphate hydrolase" negated: false notes: "Beta-lactamase class A domain with P-loop NTP hydrolase domain" - number: 2 conditions: - condition_type: InterPro id value: IPR033704 curie: InterPro:IPR033704 label: "Beta-lactamase, class A, conserved site" negated: false - condition_type: InterPro id value: IPR036157 curie: InterPro:IPR036157 label: "Beta-lactamase/transpeptidase-like superfamily" negated: false notes: "Beta-lactamase class A conserved site with transpeptidase-like superfamily" - number: 3 conditions: - condition_type: InterPro id value: IPR000629 curie: InterPro:IPR000629 label: "ATP/GTP-binding site motif A (P-loop)" negated: false - condition_type: InterPro id value: IPR001650 curie: InterPro:IPR001650 label: "Helicase, C-terminal domain" negated: false - condition_type: InterPro id value: IPR011545 curie: InterPro:IPR011545 label: "DEAD/DEAH box helicase domain" negated: false notes: "P-loop motif with helicase domains - ATP hydrolysis machinery" - number: 4 conditions: - condition_type: InterPro id value: IPR000672 curie: InterPro:IPR000672 label: "GTPase-activating protein for Ras-like GTPase" negated: false - condition_type: InterPro id value: IPR020630 curie: InterPro:IPR020630 label: "Tyrosine-protein kinase, catalytic domain" negated: false notes: "GTPase-activating protein with kinase domain" annotations: - annotation_type: KEYWORD keyword_id: KW-0378 keyword_label: "Hydrolase" category: "Molecular function" reviewed_protein_count: 0 unreviewed_protein_count: 0 created_date: "2020-05-12" modified_date: "2025-05-15" review_summary: "ARBA00022801 is an exceptionally complex rule with 4070 condition sets attempting to predict hydrolase activity using a single keyword annotation. This represents a fundamental failure in rule design with excessive complexity that is unmanageable and likely produces many false positives. The rule combines diverse hydrolase families with different mechanisms and includes domains that appear in non-hydrolytic contexts. This rule should be removed and replaced with family-specific rules." action: REMOVE action_rationale: "This rule exhibits excessive complexity (4070 condition sets vs typical <50), lacks specificity (single keyword for diverse enzyme families), includes non-hydrolytic domains (kinases, GTPases), and provides minimal functional insight. The rule is unmanageable and likely produces more false positives than useful annotations." suggested_modifications: - "Remove this rule entirely due to excessive complexity" - "Replace with family-specific rules for different hydrolase classes (e.g., serine proteases, lipases, glycosidases)" - "Use specific GO molecular function terms instead of broad keywords" - "Limit condition sets to <50 per rule with mechanistic coherence" parsimony: assessment: OVERLY_COMPLEX notes: "Rule has 4070 condition sets, which is far beyond any reasonable threshold for annotation rules. This represents extreme over-complexity that makes the rule unmanageable and likely over-fitted to training data." supported_by: - reference_id: "file:rules/arba/ARBA00022801/ARBA00022801-deep-research-manual.md" supporting_text: "This rule contains 4070 different condition sets, which is far beyond any reasonable threshold for annotation rules. Most well-designed ARBA rules have <50 condition sets. Rules with >100 condition sets are typically considered problematic. 4070 condition sets suggests systematic over-fitting." literature_support: assessment: CONTRADICTED notes: "Literature strongly suggests that broad enzyme class predictions are uninformative and that family-specific approaches are required for accurate hydrolase annotation." supported_by: - reference_id: "file:rules/arba/ARBA00022801/ARBA00022801-deep-research-manual.md" supporting_text: "Radivojac et al. (2013) shows that overly broad functional categories reduce prediction specificity. Manual curators typically use specific GO molecular function terms and avoid broad enzyme class predictions without mechanistic detail." condition_overlap: assessment: SIGNIFICANT notes: "With 4070 condition sets covering diverse hydrolase families, there is inevitable massive redundancy and conflicting domain combinations that include non-hydrolytic proteins." supported_by: - reference_id: "file:rules/arba/ARBA00022801/ARBA00022801-deep-research-manual.md" supporting_text: "Many domains in this rule appear in non-hydrolase contexts: P-loop domains present in many non-hydrolytic NTPases, protein kinases are transferases (EC 2.7) not hydrolases, GTPase-activating proteins regulate GTPases but may not be hydrolytic themselves." go_specificity: assessment: TOO_BROAD notes: "Uses only keyword KW-0378 (Hydrolase) instead of specific GO molecular function terms. Hydrolases represent ~20% of all enzymes with diverse mechanisms that cannot be adequately captured by a single keyword." supported_by: - reference_id: "file:rules/arba/ARBA00022801/ARBA00022801-deep-research-manual.md" supporting_text: "Using only a keyword annotation (KW-0378) rather than specific GO molecular function terms provides limited biological insight and reduces interoperability with other annotation systems." taxonomic_scope: assessment: TOO_BROAD notes: "Rule appears to lack taxonomic restrictions, inappropriate given that some hydrolase families are lineage-specific and enzyme nomenclature varies across domains of life." supported_by: - reference_id: "file:rules/arba/ARBA00022801/ARBA00022801-deep-research-manual.md" supporting_text: "The rule appears to lack taxonomic restrictions, which may be inappropriate given that some hydrolase families are lineage-specific, enzyme nomenclature varies across domains of life, and bacterial and eukaryotic hydrolases often have different regulatory mechanisms." confidence: 0.95 references: - id: "file:rules/arba/ARBA00022801/ARBA00022801-deep-research-manual.md" title: "Deep research analysis of ARBA00022801" findings: - statement: "Rule contains 4070 condition sets, representing excessive complexity" - statement: "Combines diverse hydrolase families with fundamentally different mechanisms" - statement: "Includes non-hydrolytic domains like kinases and GTPase regulators" - statement: "Uses overly broad keyword annotation instead of specific GO terms" - statement: "Literature supports family-specific rather than broad enzyme class predictions" - id: "PMID:23353650" title: "A large-scale evaluation of computational protein function prediction" findings: - statement: "Overly broad functional categories reduce prediction specificity" - id: "PMID:22516571" title: "Large-scale analysis of function and evolution of catalytic sites" findings: - statement: "Broad enzyme class predictions are often uninformative" supported_by: - reference_id: "file:rules/arba/ARBA00022801/ARBA00022801-deep-research-manual.md" supporting_text: "ARBA00022801 represents a systematic failure in rule design, attempting to capture excessive functional diversity with a single broad annotation. The rule's complexity (4070 condition sets) is unprecedented and unmanageable. Rather than attempting to modify this rule, it should be removed and replaced with a series of family-specific rules."