Inferred from Sequence Orthology (ISO) Evidence Code Review

Overview

ISO (Inferred from Sequence Orthology, ECO:0000266) transfers GO annotations
between orthologous genes. The code is often reliable for conserved molecular
functions, but the current reviewed corpus shows that ISO rows need a different
review question from ordinary evidence rows:

Is the source annotation itself sound, and is this specific term safe to
propagate across this orthology relationship?

The answer is not binary. Existing reviews show three common outcomes:

The transfer is correct, sometimes even redundant with target-species
experimental evidence.
The source annotation is weak, stale, circular, or miscited.
The source annotation is sound, but propagation to the target is unsafe
because the target is a paralog, a diverged family member, a different
isoform/compartment context, or a different physiological context.

Corpus Snapshot

Read-only inventory on 2026-06-29 of reviewed *-ai-review.yaml files containing
evidence_type: ISO:

Metric	Count
Reviewed ISO rows	4,189
Gene review files with ISO rows	153
Mouse ISO rows	3,228
Rat ISO rows	926
Other species ISO rows	35

Disposition across those reviewed ISO rows:

Review action	Count	Interpretation
`ACCEPT`	1,374	Conserved enough to keep as core or correct annotation.
`KEEP_AS_NON_CORE`	1,970	Often true, but contextual, tissue-specific, downstream, or generic localization.
`MARK_AS_OVER_ANNOTATED`	378	Biology may be related, but the propagated term overstates the direct role.
`REMOVE`	217	Strong evidence that the annotation should not be retained.
`MODIFY`	114	The transfer captures the right idea but needs a better GO term.
`UNDECIDED`	127	Evidence could not be adjudicated confidently.
`NEW`	9	Orthology-supported new-annotation candidates recorded in review files.

Most reviewed ISO rows are therefore not catastrophic failures. The main curation
value is discriminating correct/non-core transfers from source or propagation
defects.

Existing Review Synthesis

Calmodulin: valid family-wide transfer, with locus-specific caveats

Calm3 is the clean positive control. Mouse Calm1, Calm2, and Calm3 encode
the same calmodulin protein, so ISO transfer of core protein functions such as
calcium binding, channel
regulation, calcineurin/CaMKII-related signaling, and cytoplasmic localization is
biochemically sound. The remaining issue is not protein orthology but locus and
context: Calm3 has a distinct 3-UTR/Stau2-dependent dendritic mRNA-localization
mechanism, while many synaptic, cardiac, spindle, centrosome, and sarcomere rows
are best retained as non-core context rather than treated as defining Calm3
function.

Ghr: direct donor support mixed with circular and stale transfers

Ghr shows why ISO review must trace the donor edge. The manual donor trace
(genes/mouse/Ghr/Ghr-iso-donor-trace.md) separates
human and rat donor rows into direct receptor biology, redundant/circular
transfer chains, and stale or context-heavy rows.

Good transfers include growth hormone receptor activity, peptide hormone binding,
plasma membrane/cell surface localization, receptor complex terms, and
growth-hormone/JAK-STAT pathway terms. Problem rows include lipid binding, growth
factor binding, receptor internalization, response to estradiol, response to
cycloheximide, IGF receptor signaling, cytosol, cytoplasmic RNP granule,
proteasome-mediated catabolism, and negative regulation of GHR signaling. Several
are not simply "wrong ortholog" cases; they are transfer-of-transfer artifacts
or donor rows with no current source-side support.

The Ghr review also shows an isoform trap: extracellular-space annotations can
fit the GH-binding protein product, while signaling annotations apply to the
full-length membrane receptor.

Ang2: source and propagation defects reinforce each other

Ang2 is the strongest ISO negative control. A reproducible source trace
(genes/mouse/Ang2/Ang2-bioinformatics/RESULTS.md) found that
all 46 local ISO rows were transferred from human ANG (UniProtKB:P03950) via
GO_REF:0000119. The trace classified the human source side as:

Source-side status	Count
Current human ANG term with direct experimental support	34
Current human ANG term with inferred support only	4
No current matching human ANG source term	8

Even many directly supported human ANG annotations are unsafe for mouse Ang2,
because mouse Ang2/Angrp is a divergent angiogenin paralog: direct mouse evidence
supports RNase/tRNA-cleavage activity but not canonical angiogenesis, receptor
binding, signaling, stress-response, nuclear-trafficking, or immune-effector
roles. The reviewed Ang2 ISO rows ended up mostly REMOVE:

Action for Ang2 ISO rows	Count
`REMOVE`	41
`ACCEPT`	2
`KEEP_AS_NON_CORE`	2
`MODIFY`	1

Ang2 also exposed source-annotation hygiene problems outside the ISO block:
several experimental GOA rows trace to angiopoietin-2, angiotensin II, or ZNF418
papers rather than to mouse Ang2/Angrp. This is the clearest example where
"source annotation bad" and "propagation bad" both matter.

Broad mouse and rat ISO: the common pattern is non-core context

The largest reviewed ISO blocks are not dominated by outright removals. Many
mouse and rat signaling genes carry large ISO sets where the core molecular
function is correct but the transferred rows include many tissue, pathway,
phenotype, complex, or localization contexts. Typical review actions are
KEEP_AS_NON_CORE, MARK_AS_OVER_ANNOTATED, or MODIFY, not REMOVE.

This matters for reviewer tone: ISO is not mostly garbage. The risk is that a
large cloud of true-but-contextual annotations can obscure the small number of
source defects, stale transfers, and paralog-specific propagation failures.

Failure Taxonomy

Use two labels when reviewing IBA or ISO propagation:

A root-cause class: where the defect lives.
A biological subtype: what kind of transfer mistake it is.

Record these in review.propagation_review. Keep the narrative rationale in
review.reason; the structured object should stay mechanical.

Example:

review:
  action: REMOVE
  reason: Mouse Ang2/Angrp retains RNase activity but lacks the angiogenic function of human ANG.
  propagation_review:
    root_cause: PROPAGATION_BAD
    failure_modes:
      - WRONG_ORTHOLOG_OR_PARALOG
      - FUNCTIONAL_DIVERGENCE
    source_entities:
      - source_id: UniProtKB:P03950
        source_label: ANG
        source_status: SUPPORTS_SOURCE_BUT_NOT_TARGET
        comment: Human ANG supports angiogenesis; mouse Ang2/Angrp is non-angiogenic.

Root-cause classes

Code	Meaning	Fix target
`NO_FAILURE_CORE`	Transfer is correct and core for the target.	Keep annotation.
`NO_FAILURE_NON_CORE`	Transfer is biologically defensible but contextual or secondary.	Keep as non-core.
`SOURCE_BAD`	Source annotation is wrong, miscited, homonym-confused, or contradicted.	Fix source annotation before judging transfer.
`SOURCE_STALE_OR_MISSING`	Transferred term no longer appears on the current source record, or donor trace cannot recover it.	Remove or re-source the transfer.
`SOURCE_WEAK_OR_INFERRED`	Source exists but is only inferred, statement-level, or otherwise weak for transfer.	Downgrade confidence; require target/family corroboration.
`EVIDENCE_CIRCULAR_OR_REDUNDANT`	ISO chain transfers from another transfer, or target already has stronger direct evidence.	Do not treat ISO edge as independent support.
`PROPAGATION_BAD`	Source annotation is sound, but the term should not propagate to this target.	Fix orthology/family/node propagation.
`TERM_SCOPING_PROBLEM`	Biology is related, but the GO term is too broad, too specific, or wrong in role/qualifier.	`MODIFY` or mark over-annotated.
`UNRESOLVED`	Propagation issue was investigated but cannot yet be classified confidently.	Use `UNDECIDED` or defer until source/target evidence is resolved.

Biological subtypes

Code	Applies to	Typical signal
`WRONG_ORTHOLOG_OR_PARALOG`	ISO and IBA	Donor/source is a paralog, expanded family member, or wrong subfamily.
`FUNCTIONAL_DIVERGENCE`	ISO and IBA	Target retained fold/orthology but changed substrate, product, activity, or pathway role.
`PSEUDO_OR_SUBACTIVITY_LOSS`	Mostly IBA, sometimes ISO	Catalytic residues or a specific sub-activity are lost even though the domain remains.
`CONTEXT_OR_TISSUE_MISMATCH`	ISO and IBA	Donor evidence is tissue, developmental, organismal, or disease-context specific.
`LINEAGE_OR_TAXON_MISMATCH`	Mostly IBA	Process does not occur in the target lineage or organelle system.
`COMPARTMENT_OR_COMPLEX_MISMATCH`	ISO and IBA	Localization, complex membership, or pathway compartment does not transfer.
`REGULATORY_SIGN_INVERSION`	ISO and IBA	Family contains activators and inhibitors; positive/negative term leaks across members.
`ROLE_CONFLATION`	ISO and IBA	Substrate, regulator, effector, or specificity subunit is annotated as the agent/core machinery.
`GRANULARITY_MISMATCH`	ISO and IBA	Parent term is true but uninformative, or child term overstates specificity.
`SOURCE_MISCITATION`	ISO and IBA	Source evidence points to the wrong gene, organism, publication, or homonym.
`SOURCE_EVIDENCE_WEAK`	ISO and IBA	Source evidence is inferred, statement-level, stale, or otherwise too weak for confident propagation.
`CIRCULAR_PROPAGATION`	Mostly ISO, sometimes IBA	Propagation chain depends on another propagated annotation rather than independent source evidence.

Examples:

Ang2 human ANG-to-mouse Ang2 angiogenesis rows:
PROPAGATION_BAD + WRONG_ORTHOLOG_OR_PARALOG + FUNCTIONAL_DIVERGENCE.
Ghr receptor internalization:
SOURCE_STALE_OR_MISSING + EVIDENCE_CIRCULAR_OR_REDUNDANT.
Ghr hormone-mediated signaling:
TERM_SCOPING_PROBLEM + GRANULARITY_MISMATCH.
Calm3 calcium ion binding:
NO_FAILURE_CORE.
Calm3 spindle/centrosome contexts:
NO_FAILURE_NON_CORE + CONTEXT_OR_TISSUE_MISMATCH.

Reviewer Checklist

Use this checklist before making a strong REMOVE call on ISO or IBA.

Source annotation checked

Record the target row: gene, species, GO term, qualifier, aspect, evidence code,
reference, WITH/FROM, and assigned-by source.
Verify the GO term definition, not just the label.
Resolve the source gene/product(s) named by WITH/FROM.
Check whether the current source record still carries the same GO term.
Record the source evidence code and original source reference.
If the source evidence is experimental, verify that the cited paper supports
the source gene and term. If only abstract text is cached, avoid claiming a
curator misread full text.
Check for source-side NOT annotations or contradictory direct evidence.

Propagation checked

For ISO, confirm whether the donor-target relation is one-to-one orthology,
one-to-many, in-paralog, identical protein sequence, or broader family
similarity.
For IBA, inspect the PANTHER family/node, seed genes, WITH/FROM proteins, and
whether seeds map to the same functional subfamily as the target.
Ask whether the exact term should transfer, not only whether the broad family
function is conserved.
Check active-site residues, key domains, isoforms, compartment targeting,
lineage constraints, and target-species direct evidence when relevant.
Distinguish source defects from propagation defects. Do not hide a bad source
annotation under "orthology failure," and do not blame a source annotation when
the real problem is target-specific divergence.

Disposition recorded

Assign propagation_review.root_cause and any relevant
propagation_review.failure_modes.
Add propagation_review.source_entities entries for donor genes, seed
proteins, PANTHER nodes, family nodes, or other source entities that were
checked.
State whether the recommended fix is source-side, propagation-side, term
replacement, or non-core retention.
Prefer UNDECIDED when the source full text or family-node evidence cannot be
checked.

Action Items

[x] Review mouse Calm3 ISO annotations as the initial positive-control case.
[x] Summarize existing ISO reviews across the repository.
[x] Add an ISO/IBA failure taxonomy that distinguishes source defects from
propagation defects.
[x] Add a source/family annotation checklist for future ISO and IBA reviews.
[x] Add structured review.propagation_review schema fields for the taxonomy
and per-source entity comments.
[ ] Add a reusable ISO donor-trace script that handles multiple source entities
and writes the same source-status fields used in the Ang2 trace.