AI Target Discovery · Radioligand Therapy

Find the next great radioligand target before your competitors do.

Nuclens reads your target thesis in plain English, evaluates a catalog of oncology targets against the constraints that actually matter for radioligand therapy, and returns a ranked, fully-sourced shortlist — in minutes, not months. Every number traces back to a public database you already trust.

Rank targets free. Pay only when you generate a full decision report.

15,000+
Oncology targets, continuously refreshed
6
Public data sources, unified per target
<60s
From plain-English thesis to ranked shortlist
Built on the evidence your scientists already cite
Open TargetsUniProtHuman Protein Atlas ClinicalTrials.govDepMapPubMed
See it in action

A glimpse of the workflow.

An illustrative preview — real targets, real sourced data. Your results are driven by your own criteria.

Criteria Cell-surface targets with low kidney expression for Lu-177 therapy in prostate cancer

Click any target to expand its scorecard

1
FOLH1 PSMA
Cell-Surface Kidney: Low · 0 nTPM Tumor:Kidney 2,773× Clinical · Internalizes ✓
Sources: UniProt · Human Protein Atlas · ClinicalTrials.gov
94
Match
Cell-Surface Evidence25/25
Tumor Specificity20/20
Normal-Tissue Safety18/20
Clinical Maturity14/15
Internalization9/10
Competitive Landscape8/10
2
PSCA
Cell-SurfaceKidney: LowT:K 15,523×Clinical · 2 trials
88
Cell-Surface Evidence24/25
Tumor Specificity20/20
Normal-Tissue Safety18/20
Clinical Maturity12/15
Internalization5/10
Competitive Landscape9/10
3
STEAP1
Cell-SurfaceKidney: LowT:K 1,880×Clinical
85
Cell-Surface Evidence24/25
Tumor Specificity18/20
Normal-Tissue Safety18/20
Clinical Maturity12/15
Internalization5/10
Competitive Landscape8/10
4
TMEFF2
Cell-SurfaceKidney: LowT:K 1,921×Discovery
81
Cell-Surface Evidence23/25
Tumor Specificity18/20
Normal-Tissue Safety18/20
Clinical Maturity8/15
Internalization5/10
Competitive Landscape9/10
5
GRPR
Cell-SurfaceKidney: LowGPCR · peptide receptorDiscovery
77
Cell-Surface Evidence24/25
Tumor Specificity14/20
Normal-Tissue Safety18/20
Clinical Maturity7/15
Internalization6/10
Competitive Landscape8/10
Illustrative preview · every value shown traces to a public source Run your own — start free
Why now

The industry's biggest bets — RayzeBio ($4.1B), Fusion ($2.4B), Point Biopharma — were all races to the same scarce resource: the right target.

Radioligand therapy is the fastest-moving modality in oncology.

Approved franchises and billion-dollar acquisitions have made target selection the bottleneck everyone is fighting over.

Target triage still takes weeks of manual database work.

Localization, dosimetry-organ expression, tumor specificity, competition — pulled by hand, one tab at a time.

The teams that win won't have the most scientists — they'll have the fastest ones.

Every quarter spent hunting targets is a quarter a faster competitor spends validating them.

What Nuclens does

A radiopharma-specific reasoning engine — not a generic database.

General target-ID tools weren't built for radioligand therapy. Nuclens is. It scores every candidate on the constraints that decide whether a target can actually become a radioligand.

Ask in plain English

“Cell-surface target, low kidney expression, high tumor-to-background in pancreatic cancer.” Nuclens parses the intent — including radiopharma domain rules like alpha-therapy marrow toxicity — and filters the catalog deterministically before it ever ranks.

Scored on what matters for RLT

Cell-surface access, dose-limiting organ expression, tumor-to-kidney ratio, internalization, shedding, competitive landscape — each candidate gets a transparent scorecard, not a black-box number.

Every claim is traceable

Localization from UniProt. Expression from the Human Protein Atlas. Stage from ClinicalTrials.gov. Each data point carries its source · version · value — defensible in a program review, not just a demo.

Decision-grade reports in one click

Generate a structured target brief — biology, clinical evidence, radioligand context, safety, competition, open questions — grounded in your data and current literature. A day of desk research, on demand.

How it works

Thesis in. Ranked, sourced shortlist out.

Describe your target profile

Type the biology, indication, modality, and safety constraints in your own words — or use the hard filters.

The engine filters, then ranks

Deterministic filters fix the candidate pool; a radiopharma-tuned model scores each one across six weighted dimensions.

Review, trace, and report

Open any target's sources, compare the shortlist, and generate a full decision report when you're ready to go deep.

Built for scientists who verify

Trust comes from transparency, not confidence.

We'd rather show you the evidence and its limits than a number you can't check. That's what makes a tool safe to base a program decision on.

Sourced, versioned, verifiable

No orphan claims. Every value shown is attributed to a specific public dataset and release, so your team can open the primary source in one click.

Honest about its limits

Nuclens is a first-pass triage layer — it compresses weeks of screening into minutes. It does not replace wet-lab internalization assays, clinical dosimetry, or freedom-to-operate. AI-extracted literature is flagged for verification.

Reproducible by design

Hard filters run deterministically, so the same criteria return the same candidate pool every time — the reasoning layer only ranks within it.

Who it's for

One workflow, three decisions it accelerates.

CSO & Discovery Leadership

See the whole radioligand target landscape in an afternoon, not a quarter. Prioritize pipeline bets with the evidence attached and the reasoning you can defend to a board.

Discovery Scientists

Skip the tab-by-tab database grind. Spend your hours validating the shortlist instead of assembling it — with every source one click away.

Business Development

Map competition and white space fast. Walk into partnering conversations knowing which targets are crowded and which are genuinely open.

The next PSMA is in a database right now.

Someone is going to find it. Give your team the AI layer that gets there first — start free, and only pay when a target is worth a full report.