Dossier Registry · OB-D-001 · v1.0

The best-evidenced target in atopic dermatitis is the one nobody has drugged

If you rank every gene associated with atopic dermatitis by how much evidence supports it, and then ask which of those genes already has a drug pointed at it, do the two lists agree?

Author

Janakiram Mudumbai

Published

2026-10-24

Review status

Awaiting review

Licence

CC BY 4.0

Not yet reviewed

This dossier has been published for transparency but has not yet been signed off by the Curriculum Review Board. Treat it as a student’s working analysis, not a vetted result. When a reviewer signs it, this notice is replaced by their name and the date, and the version number increments.

What it found

They do not. Filaggrin ranks first of 4,526 associated targets, and has no approved drug and no clinical candidate in atopic dermatitis. Of the twenty best-evidenced targets, sixteen already carry an approved drug or a candidate in trials — and almost all of those sixteen act on downstream inflammation rather than on the barrier defect the genetics points to. This is the reference dossier for the Open Bench capstone: it is the standard a submission is measured against, and it is deliberately reproducible in an afternoon on a Chromebook.

Findings

The top-ranked target has no drug at all

Filaggrin (FLG) scores 0.836, first of 4,526, carrying the strongest genetic-association evidence in the set (0.952) alongside near-maximal literature support (0.983). It has zero approved drugs and zero clinical candidates in atopic dermatitis. Its small-molecule tractability is empty: filaggrin is a structural protein that forms the skin barrier, not an enzyme or a receptor with a pocket to block. Nothing here says filaggrin is a good target. It says the evidence points somewhere the industry has not gone, and the reason appears to be chemistry rather than biology.

What is drugged is mostly downstream inflammation

The targets with the heaviest drug coverage are the glucocorticoid receptor NR3C1 (28 candidates), the histamine receptor HRH1 (15), and the Janus kinases JAK1 (10) and JAK2 (7). These sit downstream of the barrier defect, in the inflammatory response that follows it. They are well drugged because they are druggable and because suppressing inflammation relieves the flare. None of them repairs the barrier.

Four of the top twenty are entirely open

FLG, IL6R, IL7R and CYP24A1 carry top-twenty evidence and have no atopic dermatitis candidate. Two of them are not obviously hard to drug: IL6R already has an approved antibody in other diseases, and CYP24A1 shows a high-quality small-molecule ligand. That is a more interesting gap than filaggrin’s, because it is harder to explain away as chemistry.

The newest candidates are moving upstream

Amlitelimab (anti-OX40L, TNFSF4) is in Phase 3 and nemolizumab (anti-IL31RA) is approved — both acting higher up than the steroid-and-antihistamine layer. The field is already moving in the direction the evidence points. This dossier does not predict that; it observes it, which is a weaker and more defensible claim.

The data

Top 20 targets, with what is already pointed at them.

Evidence score is the Open Targets overall association score. Tractability is whether that modality is plausible at all. Candidates counts drugs and clinical candidates in this disease specifically, which is why targets with drugs approved elsewhere can show zero.

#	Target	Evidence	Small molecule	Antibody	Cands.	Best stage	Example
1	FLG filaggrin	0.836	—	GO CC high conf	0	—	—
2	IL13 interleukin 13	0.747	High-Quality Pocket	Approved Drug	3	Approved	Tralokinumab
3	TYK2 tyrosine kinase 2	0.705	Approved Drug	GO CC high conf	8	Approved	Delgocitinib
4	IL4R interleukin 4 receptor	0.701	—	Approved Drug	1	Approved	Dupilumab
5	IL6R interleukin 6 receptor	0.700	Structure with Ligand	Approved Drug	0	—	—
6	JAK1 Janus kinase 1	0.662	Approved Drug	GO CC high conf	10	Approved	Abrocitinib
7	NR3C1 nuclear receptor subfamily 3 group C member 1	0.626	Approved Drug	—	28	Approved	Fluticasone Propionate
8	PDE4A phosphodiesterase 4A	0.622	Approved Drug	UniProt loc high conf	5	Approved	Crisaborole
9	PDE4D phosphodiesterase 4D	0.616	Approved Drug	UniProt loc high conf	5	Approved	Crisaborole
10	FKBP1A FKBP prolyl isomerase 1A	0.601	Approved Drug	UniProt loc med conf	3	Approved	Tacrolimus
11	JAK2 Janus kinase 2	0.601	Approved Drug	GO CC high conf	7	Approved	Delgocitinib
12	PDE4B phosphodiesterase 4B	0.600	Approved Drug	UniProt loc high conf	5	Approved	Crisaborole
13	PDE4C phosphodiesterase 4C	0.597	Approved Drug	—	5	Approved	Crisaborole
14	JAK3 Janus kinase 3	0.591	Approved Drug	GO CC high conf	6	Approved	Delgocitinib
15	HRH1 histamine receptor H1	0.590	Approved Drug	UniProt loc high conf	15	Approved	Desloratadine
16	TNFSF4 TNF superfamily member 4	0.589	—	Advanced Clinical	1	Phase 3	Amlitelimab
17	IL7R interleukin 7 receptor	0.584	—	GO CC high conf	0	—	—
18	IL22 interleukin 22	0.574	Structure with Ligand	Advanced Clinical	1	Phase 2	Fezakinumab
19	CYP24A1 cytochrome P450 family 24 subfamily A member 1	0.565	High-Quality Ligand	—	0	—	—
20	IL31RA interleukin 31 receptor A	0.560	—	Advanced Clinical	1	Approved	Nemolizumab

Shaded rows have no drug or clinical candidate in atopic eczema (atopic dermatitis).

Method

01
Resolve “atopic dermatitis” to a stable ontology identifier rather than searching by name, so the query is unambiguous and repeatable. It resolves to MONDO_0004980, labelled “atopic eczema”.
02
Retrieve the targets associated with that disease, ranked by Open Targets’ overall association score, and keep the top 30. Record each component evidence score separately — genetic association, literature, clinical, animal model, RNA expression — because the overall score hides disagreement between them.
03
Retrieve the tractability assessment for each of those targets: whether a small molecule or an antibody is a plausible modality at all.
04
Retrieve all 238 drugs and clinical candidates linked to atopic eczema, and map each one to the target or targets its mechanism of action acts on.
05
Join the three tables on gene symbol. Count candidates per target and record the most advanced clinical stage reached.
06
Read the join for disagreements between evidence rank and drug coverage. Those disagreements are the output; the ranking itself is not.

Sources

Open Targets Platform (associations, tractability, drug candidates)
Data 26.09 · API 26.9.0
MONDO Disease Ontology — atopic eczema
MONDO_0004980
ChEMBL (drug mechanisms of action, via Open Targets)
as integrated in Open Targets 26.09

Reproduce it

Do not take our word for any of this.

Paste this into the Open Targets GraphQL playground. No account, no install, no payment. It returns the association table in about two seconds. The tractability and drug-candidate queries are on the unit page for Unit 3.

Open the GraphQL playground

query AtopicDermatitisTargets {
  disease(efoId: "MONDO_0004980") {
    name
    associatedTargets(page: { index: 0, size: 30 }) {
      count
      rows {
        score
        datatypeScores { id score }
        target { id approvedSymbol approvedName }
      }
    }
  }
}

Limitations

What this cannot show.

A dossier that lists no limitations has not been read carefully enough. This section is graded as hard as the findings.

Association scores are not causal claims. Open Targets aggregates evidence that a gene and a disease are mentioned together in particular ways; a high score can reflect how heavily a gene has been studied rather than how important it is.
The literature evidence component is circular with respect to attention. Well-known genes accumulate papers, which raises their literature score, which raises their rank.
“No candidate in atopic dermatitis” is not “no candidate”. Several of these targets carry drugs approved for other diseases. The join was scoped to this one indication.
Mechanism-of-action mapping is incomplete for some candidates, so candidate counts are a floor, not an exact figure.
Clinical stage reflects the most advanced trial recorded for that indication, not whether the trial succeeded. A Phase 3 entry may have failed.
This is a database join, not an experiment. It can show where attention has and has not gone. It cannot show that an untouched target would work.

Cite this

Janakiram Mudumbai (2026). The best-evidenced target in atopic dermatitis is the one nobody has drugged. Open Bench Dossier Registry, OB-D-001 v1.0. World Peace & Prosperity Foundation. https://wppfoundation.org/registry/ob-d-001

Licensed CC BY 4.0. You may reuse, translate, and build on this, including commercially, with attribution.

Version history

v1.0 · 2026-10-24
First publication. Open Targets 26.09.

This is the capstone standard.

Eight units get you to the point where you can produce one of these yourself.

Apply to Open Bench All dossiers