Artana Bio makes rare disease investible.
We turn fragmented biology into mechanism-defined patient cohorts that support drug development, basket trials, and capital allocation.
~600
Patients across our partner foundations
150+
Partner foundations across rare neurodevelopmental disease
2M+
Cross-species genotype-phenotype associations in the evidence layer
The science exists. The economics don’t.
The biological insight to treat many rare diseases already exists. It is fragmented across literature, clinical data, and experimental systems. Each disease is treated as its own market. The math fails. Mechanism-based aggregation is what makes it work.
~200 patients with MED13 syndrome
Each ultra-rare diagnosis sits in its own silo. No investor, no payer, no pharma company will fund a $500M development program for a population that small.
$2.5M per-patient cost at that scale
Gene therapy development costs $500M–$2B. Divided across 200 ultra-rare patients, that is $2.5–7.5M each before manufacturing. The math does not work.
~$200K threshold where capital arrives
Platform approaches combined with mechanism-based clustering can reach the per-patient threshold where payers and investors engage. Only if the mechanistic groupings are scientifically rigorous.
Different diseases can share the same mechanism, and the same drug strategy.
Distinct genetic diseases converge on shared biological pathways, protein complexes, and cellular dysfunction. Group patients by mechanism instead of gene and populations aggregate, trials become feasible, and capital can be deployed.
A reasoning layer that turns fragmented biology into investible, trial-ready cohorts.
Artana Evidence is a reasoning layer built on Mondo, Monarch, ClinVar, and AlphaFold. It connects these inputs with patient phenotypes and published literature into a single layer that packages mechanistic evidence into cohorts drug developers, foundations, and capital allocators can act on. This is not another knowledge graph. It is the layer that assembles existing resources into the holistic causal evidence that FDA basket trial submissions require.
Two cohort types
Each cohort includes
- Defined patient population across diseases (size, criteria, phenotype mapping)
- Ranked mechanistic hypothesis with supporting evidence
- Linked evidence graph (literature, structural, clinical, registry data)
- Therapeutic modality recommendations
- Basket-trial design rationale aligned to FDA guidance
- IND-supporting documentation + payer evidence framework (HRU, QoL, cohort sizing)
Five-step pipeline: Normalize → Connect → Infer → Validate → Package.
Expert variant analysis takes 20–40 hours. The platform is designed to reduce this by >90%, with a precision target of ≥85% against expert ground truth (Q2 2026 benchmarking underway).
Evidence sources integrated
| Modality | Source | What it provides |
|---|---|---|
| Variant pathogenicity | ClinVar, AlphaMissense | Which variants are damaging |
| Protein structure | AlphaFold, cryo-EM | How variants disrupt binding |
| Cross-species phenotypes | Monarch KG (100+ species) | Mouse, fly, fish models that recapitulate human disease |
| Patient phenotypes | Patient registries and natural history studies | Clinical presentation across carriers |
| Literature | PubMed (40M+ citations) | Published mechanistic evidence |
| Preclinical validation | Fibroblasts, organoids, perturb-seq, mouse | Lab-tested confirmation of AI hypotheses |
Three diseases, one protein complex, ~600 patients.
Our first foundation cohort comprises three separate diagnoses caused by mutations in different subunits of the same transcriptional kinase module. Fewer than 300 patients each. No investible path individually. Published research (iScience 2022) confirms convergent CCNC/mitochondrial pathology across patient-derived cell lines, supporting cross-gene grouping. Preclinical validation is underway across multiple lab partners covering fibroblasts and iPSC, brain organoids, perturb-seq screens, and mouse phenotyping.
But the investible cohort is larger. Biomarker-defined indications based on shared mechanism can expand the addressable population to 10,000+ patients in the US, because the therapeutic target is the biology, not the gene name.
Within-gene variation in drug response may exceed between-gene variation. The right indication boundary is biomarkers, not gene names.
The pipeline beyond our first cohort
Our first cohort is one of many. Two expansion axes: within-mechanism (biomarker-defined indication grows the addressable population from ~600 to 10,000+) and across-mechanisms (the founding kinase module has 26 core subunits; chromatin and transcription regulators are the next nodal cluster). The same mechanism-based grouping extends across monogenic rare disease.
Biomarker-defined indications are FDA precedent today.
Keytruda’s 2017 tumor-agnostic approval established that a biomarker, not a tissue type, can define an indication. The same logic applies to rare disease: a shared mechanism, validated by a companion diagnostic, can define a cohort across gene boundaries. Recent FDA draft guidance and PRV reauthorization make the path stronger, not the foundation.
Evidence assets now. Platform at scale.
Near-term, we sell evidence assets: mechanistic variant reports, cohort analyses, and diligence packages. Each engagement generates data that trains the platform. By 2028, when the evidence base is large enough, R&D teams subscribe to run their own analyses. The services prove the platform works. The platform captures the margin. Every engagement makes the next one more precise. That is the compounding asset that distinguishes us from a consulting firm.
Built by the team that has already assembled the data, patients, and infrastructure.
Three of us built rare disease knowledge systems together at CZI. Stanford clinical data infrastructure expertise. Patient-founder lived experience. Scientific Advisory Board across organoids, nodal biology, structural biology, and clinical genetics. Specific advisor and partner names are shared in fundraising and consulting conversations. Get in touch.
Built for regulated, high-stakes science.
Enterprise security isn't a feature tier. It's a design requirement. Artana Bio is architected from the ground up for institutions where access control and data integrity are non-negotiable.
Private by default
Data is never shared, indexed, or exposed across tenants. Isolation is enforced at the architecture level. Not governed by policy alone.
Role-based access controls
Define granular permissions per team member, project, and data class. Access is least-privilege by default and auditable at every layer.
Audit logging
Every query, curation action, and permission change is logged with full user attribution and timestamps. Ready for institutional review.
Encrypted data at rest and in transit
AES-256 at rest. TLS 1.3 in transit. Encryption keys are managed per-customer with configurable residency controls.
Enterprise deployment options
Deploy to your own cloud environment, on-premises infrastructure, or a dedicated hosted tenant. Full data residency control included.
Precision medicine becomes viable when cohorts become investible.
We are not a drug company. We are not a CRO. We are not a database. We are the infrastructure layer that makes mechanism-based medicine actionable.