Protein Dossier — THBS4 (Thrombospondin-4)
MR feasibility tier: A — Published pQTL-MR estimates exist for this protein (retrieved below - not computed here).
1. Published MR estimates (retrieved, not computed)
| Outcome |
beta |
se |
p |
method |
nSNP |
cis/trans |
coloc |
| Non-cancer illness code self-reported: sleep apnoea |
0.48 |
0.158 |
0.00234 |
Wald ratio |
1 |
cis |
NA |
| Diagnoses - main ICD10: K40 Inguinal hernia |
0.173 |
0.071 |
0.015 |
Wald ratio |
1 |
cis |
NA |
| Non-cancer illness code self-reported: gastro-oesophageal reflux (gord) or gastric reflux |
0.137 |
0.0578 |
0.018 |
Wald ratio |
1 |
cis |
NA |
| Caudate volume |
-67 |
28.8 |
0.0199 |
Wald ratio |
1 |
cis |
NA |
| Non-cancer illness code self-reported: diverticular disease or diverticulitis |
0.229 |
0.105 |
0.0286 |
Wald ratio |
1 |
cis |
NA |
| Eye problems or disorders: Injury or trauma resulting in loss of vision |
0.289 |
0.138 |
0.0361 |
Wald ratio |
1 |
cis |
NA |
| Non-cancer illness code self-reported: bladder problem (not cancer) |
0.281 |
0.139 |
0.0432 |
Wald ratio |
1 |
cis |
NA |
| Squamous cell lung cancer |
-0.299 |
0.149 |
0.0448 |
Wald ratio |
1 |
cis |
NA |
| Non-cancer illness code self-reported: enlarged prostate |
-0.329 |
0.169 |
0.0512 |
Wald ratio |
1 |
cis |
NA |
| Diagnoses - main ICD10: L03 Cellulitis |
0.238 |
0.123 |
0.0525 |
Wald ratio |
1 |
cis |
NA |
| Femoral neck bone mineral density |
-0.0821 |
0.0425 |
0.0532 |
Wald ratio |
1 |
cis |
NA |
| Thalamus volume |
-70.5 |
36.8 |
0.0554 |
Wald ratio |
1 |
cis |
NA |
| …and 66 more outcomes (see JSON) |
|
|
|
|
|
|
|
2. pQTL instrument availability (Tier-B probe)
| Dataset |
Trait |
Author |
Year |
prot-c-3340_53_1 |
TSP4 |
Suhre K |
2019 |
3. GWAS Catalog results — traits with signal at this locus
49 association rows across 31 traits (45 genome-wide significant rows). Associations are loci, not causal claims; the mapped gene at a locus is not necessarily the effector gene.
| Trait |
best p |
lead SNP |
n assoc |
study |
MR status |
| COMP/THBS4 protein level ratio |
2e-435 |
rs2438637 |
1 |
GCST90314193 |
no MR -> candidate analysis |
| FAP/THBS4 protein level ratio |
2e-320 |
rs2438637 |
1 |
GCST90314785 |
no MR -> candidate analysis |
| Circulating THBS4 levels |
3e-267 |
rs35351529 |
4 |
GCST90860466 |
no MR -> candidate analysis |
| Thrombospondin-4 levels |
8e-110 |
rs13167730 |
3 |
GCST90249997 |
no MR -> candidate analysis |
| Cerebrospinal fluid protein THBS4 levels |
3e-55 |
rs256438 |
1 |
GCST90945058 |
no MR -> candidate analysis |
| THBS4 protein levels |
3e-27 |
rs568927878 |
6 |
GCST90470855 |
no MR -> candidate analysis |
| Height |
4e-27 |
rs2451933 |
1 |
GCST90245848 |
no MR -> candidate analysis |
| Smoking initiation |
1e-22 |
rs7707036 |
2 |
GCST90243985 |
no MR -> candidate analysis |
| Corneal curvature |
2e-22 |
rs13180294 |
3 |
GCST90012795 |
no MR -> candidate analysis |
| Corneal resistance factor (MTAG) |
2e-19 |
rs13167730 |
2 |
GCST90102517 |
no MR -> candidate analysis |
| Impedance of arm right (UKB data field 23109) |
3e-19 |
rs41272276 |
1 |
GCST90468172 |
no MR -> candidate analysis |
| Refractive error |
1e-18 |
rs256438 |
3 |
GCST90841196 |
no MR -> candidate analysis |
| …and 19 more traits (see JSON) |
|
|
|
|
|
4. Phenome map — where this gene is a genetic locus, vs. where MR exists
Top diseases by Open Targets association (of 535 total). Associations are loci, not causal claims. The causal-status column is a four-state triage per pair: established (curated) = a curated clinical assertion exists (ClinGen/G2P/GEL/Orphanet/ClinVar — any validity level, MR adds little); exploratory rare-variant signal = ExWAS burden evidence without curation — a candidate NEW gene-disease relationship; common-variant locus = GWAS signal, classic pQTL-MR territory; multi-layer = burden+GWAS together, an allelic-series candidate (the strongest causal setup). Burden estimand is carrier-vs-noncarrier, not per-SD MR.
| Disease |
genetic assoc. |
burden (ExWAS) |
causal status |
MR status |
| Abnormality of the skeletal system |
0.894 |
0.799 |
multi-layer: burden+GWAS (allelic-series candidate) |
no MR -> candidate analysis |
| Abnormality of refraction |
0.473 |
— |
common-variant locus |
no MR -> candidate analysis |
| attention deficit-hyperactivity disorder |
0.339 |
— |
common-variant locus |
no MR -> candidate analysis |
| substance abuse |
0.339 |
— |
common-variant locus |
no MR -> candidate analysis |
| smoking initiation |
0.284 |
— |
common-variant locus |
no MR -> candidate analysis |
| Varicose veins |
0.166 |
— |
common-variant locus |
MR: beta=0.136, p=0.108 (cis) |
| post term pregnancy |
0.15 |
— |
common-variant locus |
no MR -> candidate analysis |
| benign thyroid gland neoplasm |
0.146 |
— |
common-variant locus |
no MR -> candidate analysis |
| vein disorder |
0.129 |
— |
common-variant locus |
no MR -> candidate analysis |
| lymphatic system disorder |
0.129 |
— |
common-variant locus |
no MR -> candidate analysis |
| glaucoma |
0.122 |
— |
common-variant locus |
no MR -> candidate analysis |
Of the 11 rows above, 10 have no MR estimate in this resource. Across all retrieved diseases for this gene: 0 exploratory rare-variant signal(s), 1 multi-layer (allelic-series candidate) pair(s). Final triage still belongs to a statistical geneticist.
5. Downstream annotation (druggability & safety preview)
| Layer |
Result |
| ChEMBL druggability |
not available — no ChEMBL target (undrugged) |
| gnomAD constraint |
pLI=6.7e-24, LOEUF=0.927 — LoF-tolerant |
| GWAS Catalog |
62 unique SNPs / 115 rows |
| ClinVar |
158 records; 0 pathogenic in sample of 30 |
| PharmGKB/ClinPGx |
no annotations |
phenome — Top 30 of 535 associated diseases by overall score. genetic_association aggregates GWAS common-variant AND rare-variant evidence. These are ASSOCIATIONS (loci), not causal claims.chembl — No ChEMBL target for ‘THBS4’.clinvar — Pathogenic count is over the 30 record(s) retrieved, NOT over all 158 ClinVar records for this gene; it is a sample, not a rate.pharmgkb — No PharmGKB/ClinPGx clinical annotations (gene may not be a pharmacogene).gwas_traits — Top 20 of 31 traits by best p-value, aggregated from 49 association rows. These are GWAS ASSOCIATIONS (loci), not causal claims; mapped genes at a locus are not necessarily the effector gene.
Sources
uniprot: https://www.uniprot.org/uniprotkb/P35443 — UniProt release 2026_02 (10-June-2026)mr_outcomes: https://epigraphdb.org/pqtl/ — EpiGraphDB pQTL MR (Zheng et al., Nat Genet 2020) — pre-computed two-sample MR; retrieved, not computed by this agent; EpiGraphDB build 1.0, pQTL dataset v3.0phenome: https://platform.opentargets.org/target/ENSG00000113296/associations — Open Targets data release 26.06gnomad: https://gnomad.broadinstitute.org/gene/THBS4 — gnomAD constraint via GraphQL API (reference genome GRCh38)gwas: https://www.ebi.ac.uk/gwas/genes/THBS4 — GWAS Catalog REST (live; release not exposed by this endpoint)clinvar: https://www.ncbi.nlm.nih.gov/clinvar/?term=THBS4%5Bgene%5D — ClinVar build Build260809-1055.1gwas_traits: https://www.ebi.ac.uk/gwas/genes/THBS4 — GWAS Catalog search API (live; release not exposed)
Provenance
- Generated: 2026-08-14T05:20:46 · Tier: A
- Fully mechanical: every cell above is rendered from tool return values. No language model wrote any part of this dossier.
- MR estimates, where present, are retrieved from published work (EpiGraphDB pQTL, Zheng et al. Nat Genet 2020); nothing is computed here.
- Tool errors this run: none