View original ARBA rule on UniProt
A massive rule with 1094 condition sets that attempts to annotate the entire transferase enzyme class (EC 2) using GO:0016740 (transferase activity). The rule encompasses diverse transferase families including glycosyltransferases, acetyltransferases, protein kinases, ubiquitin-conjugating enzymes, and others, but also erroneously includes non-transferase domains like PAS signaling domains.
Condition-set counts describe the sets recorded in this review, which may omit the full rule.
ARBA00027723 represents a fundamentally flawed approach to transferase annotation that attempts to capture the entire EC 2 enzyme class in a single rule. With 1094 condition sets spanning 406 InterPro domains and 73 PANTHER families, this rule is impossibly complex and unmaintainable. Critical analysis reveals the inclusion of non-transferase domains (IPR000014 PAS domain) alongside legitimate but highly diverse transferase families. The use of GO:0016740 "transferase activity" as the target annotation loses essential functional specificity, conflating enzymes with vastly different biological roles - from metabolic glycosyltransferases to regulatory protein kinases to protein degradation enzymes. The rule violates basic annotation principles by trading specificity for comprehensiveness, resulting in annotations that provide minimal biological insight. The correct approach would be to replace this mega-rule with focused, family-specific rules that use appropriate child GO terms and maintain biological coherence within manageable complexity limits.
This rule exhibits multiple critical flaws that make it unsuitable for accurate protein annotation: (1) Excessive complexity with 1094 condition sets that exceeds maintainability and validation limits by two orders of magnitude, (2) Inappropriate inclusion of non-transferase domains (e.g., PAS signaling domains), (3) Overly broad GO term that loses functionally critical specificity needed for meaningful biological annotation, (4) Biological incoherence by conflating highly diverse enzyme families with different substrates, mechanisms, and biological roles, and (5) Violation of annotation best practices by creating a "mega-rule" instead of focused, family-specific rules.
The rule has 1094 condition sets, which is approximately 90x the recommended maximum of 12 condition sets. This creates an unmaintainable and unvalidatable rule that cannot be properly reviewed or maintained by curators. Analysis of the first few condition sets reveals a mix of legitimate transferase domains alongside inappropriate inclusions like PAS signaling domains.
While individual transferase families are well-characterized in the literature, there is no biological rationale for unifying all EC 2 transferases under a single broad annotation. The enzyme classification system specifically maintains detailed subclassification because substrate specificity and biological context are functionally critical. The inclusion of non-transferase domains contradicts established biochemical knowledge.
With 1094 condition sets attempting to cover all transferases, there is guaranteed to be extensive redundancy and complete functional overlap between many condition sets targeting the same protein families. The scale makes meaningful overlap analysis impossible.
GO:0016740 "transferase activity" is a high-level molecular function term that provides minimal biological insight. More specific child terms exist for all major transferase subclasses and should be used instead. This term is appropriate only when the specific transfer reaction cannot be determined.
The rule lacks taxonomic restrictions and would apply to all domains of life, but many of the included enzyme families have lineage-specific distributions and functions that would be better served by targeted rules.
Contains 1094 condition sets with 406 InterPro domains and 73 PANTHER families targeting GO:0016740
GO:0016740 represents the broad transferase activity class encompassing all EC 2 enzymes
id: ARBA00027723
description: >-
A massive rule with 1094 condition sets that attempts to annotate the entire
transferase enzyme class (EC 2) using GO:0016740 (transferase activity).
The rule encompasses diverse transferase families including glycosyltransferases,
acetyltransferases, protein kinases, ubiquitin-conjugating enzymes, and others,
but also erroneously includes non-transferase domains like PAS signaling domains.
status: COMPLETE
rule_type: ARBA
rule:
rule_id: ARBA00027723
condition_sets: [] # Unable to analyze due to excessive complexity (1094 condition sets)
go_annotations:
- go_id: GO:0016740
go_label: transferase activity
aspect: MF
reviewed_protein_count: 0
unreviewed_protein_count: 0
created_date: null
modified_date: null
entries: []
review_summary: >-
ARBA00027723 represents a fundamentally flawed approach to transferase annotation
that attempts to capture the entire EC 2 enzyme class in a single rule. With
1094 condition sets spanning 406 InterPro domains and 73 PANTHER families, this
rule is impossibly complex and unmaintainable. Critical analysis reveals the
inclusion of non-transferase domains (IPR000014 PAS domain) alongside legitimate
but highly diverse transferase families. The use of GO:0016740 "transferase activity"
as the target annotation loses essential functional specificity, conflating enzymes
with vastly different biological roles - from metabolic glycosyltransferases to
regulatory protein kinases to protein degradation enzymes. The rule violates basic
annotation principles by trading specificity for comprehensiveness, resulting in
annotations that provide minimal biological insight. The correct approach would
be to replace this mega-rule with focused, family-specific rules that use
appropriate child GO terms and maintain biological coherence within manageable
complexity limits.
action: DEPRECATE
action_rationale: >-
This rule exhibits multiple critical flaws that make it unsuitable for accurate
protein annotation: (1) Excessive complexity with 1094 condition sets that
exceeds maintainability and validation limits by two orders of magnitude,
(2) Inappropriate inclusion of non-transferase domains (e.g., PAS signaling domains),
(3) Overly broad GO term that loses functionally critical specificity needed
for meaningful biological annotation, (4) Biological incoherence by conflating
highly diverse enzyme families with different substrates, mechanisms, and
biological roles, and (5) Violation of annotation best practices by creating
a "mega-rule" instead of focused, family-specific rules.
suggested_modifications:
- Remove this rule entirely and replace with focused, family-specific rules
- Create separate rules for major transferase classes using specific child GO terms
- Implement proper complexity limits to prevent creation of similar mega-rules
- Ensure all included domains are actually relevant to the target GO annotation
parsimony:
assessment: OVERLY_COMPLEX
notes: >-
The rule has 1094 condition sets, which is approximately 90x the recommended
maximum of 12 condition sets. This creates an unmaintainable and unvalidatable
rule that cannot be properly reviewed or maintained by curators. Analysis
of the first few condition sets reveals a mix of legitimate transferase domains
alongside inappropriate inclusions like PAS signaling domains.
supported_by:
- reference_id: "file:rules/arba/ARBA00027723/ARBA00027723.json"
supporting_text: "Analysis shows 1094 condition sets with 406 unique InterPro domains and 73 PANTHER families"
literature_support:
assessment: WEAK
notes: >-
While individual transferase families are well-characterized in the literature,
there is no biological rationale for unifying all EC 2 transferases under a
single broad annotation. The enzyme classification system specifically maintains
detailed subclassification because substrate specificity and biological context
are functionally critical. The inclusion of non-transferase domains contradicts
established biochemical knowledge.
supported_by:
- reference_id: "external:EC_classification"
supporting_text: "EC class 2 transferases are divided into 10 subclasses based on the type of group transferred, reflecting functional diversity that requires specific annotation"
condition_overlap:
assessment: COMPLETE
notes: >-
With 1094 condition sets attempting to cover all transferases, there is
guaranteed to be extensive redundancy and complete functional overlap
between many condition sets targeting the same protein families. The scale
makes meaningful overlap analysis impossible.
go_specificity:
assessment: TOO_BROAD
notes: >-
GO:0016740 "transferase activity" is a high-level molecular function term
that provides minimal biological insight. More specific child terms exist
for all major transferase subclasses and should be used instead. This term
is appropriate only when the specific transfer reaction cannot be determined.
supported_by:
- reference_id: "external:QuickGO"
supporting_text: "GO:0016740 definition: 'Catalysis of the transfer of a group...from one compound to another. Transferase is the systematic name for any enzyme of EC class 2.'"
taxonomic_scope:
assessment: TOO_BROAD
notes: >-
The rule lacks taxonomic restrictions and would apply to all domains of life,
but many of the included enzyme families have lineage-specific distributions
and functions that would be better served by targeted rules.
confidence: 0.95
references:
- id: file:rules/arba/ARBA00027723/ARBA00027723.json
title: Original ARBA rule definition
findings:
- statement: Contains 1094 condition sets with 406 InterPro domains and 73 PANTHER families targeting GO:0016740
- id: external:QuickGO
title: GO term definition lookup
findings:
- statement: GO:0016740 represents the broad transferase activity class encompassing all EC 2 enzymes