Zebrafish Paralog Pair Comparisons

Zebrafish Paralog Pair Comparisons

Back to project

Each pair gets a folder pairs/<geneA>_<geneB>/ containing:

The two gene reviews themselves live as usual in genes/DANRE/<gene>/.

Pairs

Pair PANTHER call Fate (summary)
cryaba / cryabb TGD_or_lineage MIXED: uneven expression split (cryaba lens-restricted in adults, cryabb broad); both keep chaperone activity with different strengths; shared dose-sensitive lens role. TGD origin not settled
col1a1a / col1a1b TGD_tree DOSAGE with a minor expression partition: col1a1b (alpha3(I)) is a co-expressed, partly interchangeable chain of the same type I trimer; col1a1a is dose-limiting and alone makes fin-fold actinotrichia; no protein-level innovation shown
mitfa / mitfb TGD_likely_parallel PARTITION (expression); protein conserved (mitfb rescues nacre); residual redundancy in xanthophores and RPE; resolved differently in medaka
pax6a / pax6b TGD_likely_parallel PARTITION (expression) on a shared dose-sensitive core: pancreas pax6b-only, habenula pax6a-only, eye shared; equal transactivation
sox9a / sox9b TGD_or_lineage PARTITION (expression); sox9b-only roles are ancestral SOX9 roles; redundancy only in shared domains
elna / elnb unresolved (TGD per literature) MIXED: expression partition plus protein-level innovation in elnb (bulbus arteriosus stiffness; elna and Polypterus eln fail to rescue)
fgf8a / fgf8b TGD_tree_no_gar PARTITION (expression), lopsided: fgf8a keeps gastrula, heart, telencephalon, pharyngeal and fin AER domains; fgf8b shares only MHB, somite and ear; protein equivalent in injection assays; no fgf8b loss-of-function data; stickleback split the heart domain the other way
grk7a / grk7b TGD_tree MIXED: asymmetric expression partition (grk7a pan-cone and dominant in larvae; grk7b enriched in adult UV cones); both keep rhodopsin kinase activity, grk7b reportedly >30-fold slower; grk7b backs up grk7a only in the pineal
tbx5a / tbx5b TGD_likely_parallel MIXED: both copies essential and non-redundant (heart looping, fin); fin role split (tbx5a initiation, tbx5b outgrowth); one-sided expression loss in tbx5b; protein divergence inferred (49% identity, no reciprocal mRNA cross-rescue), not established
vcla / vclb TGD_tree MIXED: conserved protein with expression-biased partition (vclb-only coronary/epicardial role, lethal alone; vcla dispensable even maternal-zygotic); shared dose-sensitive endothelial barrier role; vclb upregulation in vcla mutants is mRNA-decay-triggered adaptation, no protein-level compensation shown

Random sample (batches 3-4)

Drawn at random (seed 20260928) from the 778 PANTHER TGD_tree 1:1 pairs. From batch 4 on, a
pair also had to be confirmed by Ensembl Compara (duplication node at Teleostei,
Osteoglossocephalai or Clupeocephala); see random_sample.tsv and the
summary table. Four batch 3 pairs failed that check but are kept
here as reviewed pairs: eef1da/eef1db, olfm3a/olfm3b, si:dkey-283b1.7/vwc2 and
magi3b/wu:fi36a10. vcla/vclb was drawn but had already been reviewed in batch 2, so it is not
repeated in this table. Every page carries a Sample record line.

Pair Fate Level Evidence Identity Summary
smad3a / smad3b BACKUP (provisional) both experimental_both 94.1% Conserved R-SMADs acting alike in side-by-side assays and broadly co-expressed; only the double knockout has a reported phenotype; single-mutant data not visible
eef1da / eef1db UNRESOLVED none expression_only 40.5% (long isoforms; GEF region 89%) No functional data; both keep an intact GEF domain and are co-expressed in all Bgee tissues; eef1da is maternal and early-embryo biased
si:dkey-283b1.7 / vwc2 UNRESOLVED (TGD not confirmed) protein experimental_one 32.6% vwc2 is the VWC2 ortholog (secreted BMP antagonist, forebrain; morpholino data); si:dkey-283b1.7 is an unstudied, fast-evolving brorin-family gene; Ensembl Compara and gar synteny do not support the PANTHER pairing
magi3b (LOC564220) / wu:fi36a10 UNRESOLVED, leaning non-redundant expression experimental_both 58.2% MAGI3 co-orthologs with a conserved scaffold; each F0 crispant impairs brain vascularization and the double is no worse; wu:fi36a10 broad and gar-like, magi3b narrower; Compara dates the duplication to Euteleostomi, not the TGD
tmc2a / tmc2b PARTITION expression experimental_both 68.0% Lopsided, overlapping zones: tmc2b dominates lateral line and peripheral ear, tmc2a the central ear and saccule; double mutants worse than singles; Tmc2b transgene rescues all organs, Tmc2a untested; TGD only partly supported by synteny
lhfpl5a / lhfpl5b PARTITION expression experimental_both 76.5% Clean organ split: lhfpl5a inner ear only, lhfpl5b lateral line only; each mutant loses transduction only in its organ; Lhfpl5a rescues lhfpl5b mutants; double-conserved synteny
abi1a / abi1b UNRESOLVED expression expression_only 79.2% Conserved protein; timing split (abi1b maternal/cleavage, abi1a zygotic and broad like gar ABI1); ancestral maternal loading unknown; only a mosaic F0 abi1a knockdown, which has no heart phenotype
olfm3a / olfm3b UNRESOLVED expression expression_only 83.2% Conserved protein; olfm3a is the larval brain/retina copy, olfm3b near-silent in larvae with weak adult calls; consistent with partition or with olfm3b decline; Ensembl dates the duplication to Gnathostomata
agxta / agxtb PARTITION protein sequence_only 62.4% Targeting signals split: agxta keeps only a C-terminal peroxisomal-type signal, agxtb only the gar-like N-terminal mitochondrial sequence (gar and tetrapods have both); catalytic residues and liver/kidney expression shared; pattern recurs across teleosts; untested experimentally
slc7a10a / slc7a10b UNRESOLVED expression experimental_one 77.8% Protein conserved (all tested Asc-1 residues identical); slc7a10a broad, slc7a10b narrow and closer to gar; only slc7a10b has a mutant (diet-induced weight gain); ancestral tissue pattern uncertain
exoc3l2a / exoc3l2b UNRESOLVED both expression_only 50.9% EXOC3L2 co-orthologs (PANTHER's EXOC3L4 label is wrong); exoc3l2a endothelial, exoc3l2b neural crest/arches in a separate study, never compared side by side; exoc3l2a evolves about 2x faster; double-conserved synteny; no functional data
samsn1a / samsn1b UNRESOLVED expression expression_only 37.8% Diverged SH3-SAM adaptor with a conserved SH3 domain; in situ records suggest separate domains (samsn1a blood, macrophages, lens; samsn1b pineal, photoreceptors) but come from different screens, and bulk data overlap; medaka keeps only one copy
ctdspl2a / ctdspl2b UNRESOLVED (consistent with BACKUP or DOSAGE) none expression_only 70.9% Both keep the catalytic DLDET motif and a near-identical phosphatase domain; both maternal and ubiquitous like gar, ctdspl2a 2-3x higher; medaka keeps both copies; no functional data
mapre3a / mapre3b UNRESOLVED expression expression_only 83.0% Conserved EB3 plus-end tracking protein; mapre3b maternal and broad like gar MAPRE3, mapre3a zygotic and highest in retina/brain/testis; could be a partition or a declining mapre3a; no functional data; weak synteny support
sh3glb2a / sh3glb2b UNRESOLVED none expression_only 73.6% Conserved endophilin-B2 domains in both; co-expressed in every adult tissue like gar, sh3glb2b dominant maternally; double-conserved synteny; backup, dosage or quantitative partition indistinguishable; PANTHER misplaces sh3glb2a in a drebrin-like subfamily
guca1c / guca1d UNRESOLVED, leaning protein-level divergence protein experimental_both 75.5% GCAP3 co-orthologs, both cone-specific in all cone types (guca1c about 10x higher); both activate the cyclase but at very different Ca2+ (about 30 vs 400 nM); ancestral state unassayed; no guca1d mutant
prom1a / prom1b PARTITION expression experimental_both 54.4% Lopsided: prom1a keeps the broad, maternal, gar-like pattern; prom1b is retina/brain-restricted and the prominin photoreceptors depend on (only prom1b mutants lose outer segments); complementary in adult brain; double-conserved synteny
tp53bp2a / tp53bp2b UNRESOLVED expression experimental_both 60.9% Both copies act alike in the one side-by-side study (Irs-1 binding, Akt inhibition; morpholino and overexpression only) and keep the p53-binding module; timing split (tp53bp2b near background during gastrula-pharyngula); weak synteny
tusc2a / tusc2b UNRESOLVED expression expression_only 80.2% No zebrafish literature; myristoylation site and calcium-binding motif conserved in both; tusc2a maternal-biased, tusc2b the main zygotic and adult copy; double-conserved synteny through gar
gad1a / gad1b UNRESOLVED expression experimental_one 86.3% Same enzyme in both copies; gad1b is the pan-GABAergic copy (~20x more abundant; null mutant has seizure-like activity); gad1a weakly expressed; the classic 'GAD67' probe is gad1b by sequence, so some 'gad1a' expression records may be gad1b signal
gria1a / gria1b UNRESOLVED both expression_only 82.5% Channel core intact in both; largely overlapping neural expression; gria1b evolves faster (also in medaka) and lacks the S831/S845-equivalent regulatory serines; no copy-resolved functional data

Template

---
title: "<geneA> / <geneB>"
autolink_gene_symbols: false
---

# <geneA> / <geneB>

[Back to pairs](../README.html)

**Bottom line:** <one or two sentences: the fate classification and the key evidence>

| | <geneA> | <geneB> |
|---|---|---|
| UniProt | | |
| Human ortholog | | |
| Review | [genes/DANRE/<geneA>](...) | ... |

## 1. Evidence the pair comes from the TGD
PANTHER call and evidence (row in panther_tgd_pairs.tsv: branch, gar/medaka
co-orthologs); synteny or phylogenetic papers; state what is and is not established.

## 2. Protein-level comparison
Identity (from annotation-comparison.md); domain architecture; conservation of
key functional residues/motifs; any biochemical comparison (e.g. activity assays,
cross-rescue). Does each copy retain the ancestral molecular function?

## 3. Expression
Published expression domains of each copy, and of the ancestral (pre-TGD,
e.g. gar) or tetrapod single-copy gene where known. Shared vs copy-specific domains.

## 4. Experimental evidence of function
Per copy: mutants (note the allele type: PTC/NMD vs RNA-less), morphants,
double mutants, rescue and cross-rescue, compensation by the paralog.

## 5. Fate classification
One of: PARTITION (subfunctionalization) / INNOVATION (neofunctionalization) /
BACKUP (redundancy) / DOSAGE / MIXED / UNRESOLVED. Give the level of the claim
(expression vs protein), the evidence for it, confidence, and what would change it.

## 6. GO annotation consistency across the pair
Summarize annotation-comparison.md: asymmetries that are biologically justified
(expression partition, protein divergence) vs artefacts of which copy was
studied; annotations that should be copy-specific; annotations missing from one
copy; how IBA/ISO propagation treats the pair.

## 7. Open questions

## References
Every factual claim carries [PMID:N "verbatim quote"] (checked with
scripts/check_quotes.py) or points to a file.