Snapshots, not trajectories
Static omics describe where a disease is, not where it’s going or how it responds to intervention.
Computational pathobiology · Closed-loop translational discovery
Simulate the disease. Prove it at the bench.
Static omics and standalone prediction tools can’t explain a disease in motion. Diseases AI builds multi-scale cellular world models, then closes the loop through automated biofoundries and patient digital twins, so the targets you advance are the ones that survive the clinic.
-
01 · Model
Simulate disease dynamics in real time
-
02 · Discover
Expose hidden drivers and vulnerabilities
-
03 · De-risk
Validate before the clinic does it for you
The 2026 shift
Disease is a moving system. Stop modeling it as a snapshot.
Structure prediction solved one step. Attrition lives in the rest: heterogeneous cell states, shifting networks, and the gap between a computational hit and evidence a clinician trusts. We build for that gap, as one continuous loop.
Tools, not loops
Isolated models generate candidates. Without experimental feedback, they never learn what failed or why.
A prediction becomes evidence only after it survives perturbation: in silico, in the biofoundry, and in the patient digital twin.
Three pillars
From cellular world models to de-risked translation
One computational pathobiology stack, built for researchers who need mechanism, not just ranked lists.
Pillar 01
Dynamic Disease Modeling
Beyond static omics: multi-scale cellular world models that simulate disease progression and cellular perturbation in real time.
- Molecular to tissue scale in a single model
- Predict genetic and chemical perturbations before the bench
- Retrained by every experimental result
Snapshot → trajectory
Pillar 02
Complex Target & Pathway Discovery
Uncover hidden drivers, synthetic lethality and network vulnerabilities in heterogeneous, multi-factorial disease.
- Causal network inference across patient subtypes
- Context-specific synthetic-lethal dependencies
- Targets that conventional screens overlook
Correlation → causation
Pillar 03
Translational De-risking
Bridge in silico discovery with automated wet-lab biofoundries and patient digital twins to cut clinical attrition.
- Biofoundries test hypotheses at scale, autonomously
- Digital twins stress-test efficacy and stratification
- Pre-specified go / no-go gates at every stage
Candidate → conviction
The platform
One closed loop from simulation to patient
Models propose, biofoundries test, digital twins stress-test. Every result retrains the world model that chose the next experiment.
-
Dry lab
01
Model
Fuse multi-omic, imaging and longitudinal data into multi-scale cellular world models of disease.
-
Dry lab
02
Discover
Run in silico perturbations to rank drivers, synthetic-lethal pairs and network vulnerabilities.
-
Wet lab
03
Validate
Automated biofoundries run the highest-information experiments against pre-specified acceptance criteria.
-
Clinic
04
Translate
Stress-test efficacy and stratification in patient digital twins before committing to trial design.
Continuous feedback: biofoundry and clinical-cohort results retrain the world model and sharpen the next round of hypotheses.
Pipeline & research
From first signal to clinical readout
Select a stage to filter our therapeutic verticals. Every program moves through the same gated path.
Showing all 6 programs
-
Oncology Discovery
Resistance mechanisms in solid tumors
Mapping adaptive-resistance signatures to nominate combination targets.
Stage 1 of 3: DiscoverySmall molecule
-
Rare genetic disease Discovery
Variant-to-function mapping
Linking undiagnosed variants to functional readouts to reveal tractable pathways.
Stage 1 of 3: DiscoveryGenetic medicine
-
Neuroinflammation Pre-clinical
Glial-state modulation
Validating state-specific targets in patient-derived cellular models.
Stage 2 of 3: Pre-clinicalBiologic
-
Fibrosis & metabolic Pre-clinical
Organ-fibrosis driver
Lead optimization with efficacy and exposure biomarkers in ex vivo tissue.
Stage 2 of 3: Pre-clinicalSmall molecule
-
Autoimmune disease Phase I/II
Biomarker-stratified cohorts
Patient stratification to inform dose selection and response endpoints.
Stage 3 of 3: Phase I/IIBiologic
-
Infectious disease Phase I/II
Host-response signatures
Host-response readouts used to define trial endpoints and responder groups.
Stage 3 of 3: Phase I/IISmall molecule
Programs are investigational and have not been approved by any regulatory authority for clinical use.
Scientific rigor
Evidence, not enthusiasm
Every program is held to the same standards and passes the same validation gates before it advances.
Orthogonal
Findings are confirmed across independent assay modalities, not a single readout.
Pre-specified
Analysis plans and go / no-go criteria are defined before data is generated.
Blinded
Key readouts are scored without knowledge of treatment condition.
Reproducible
Versioned data, code and protocols so every result can be re-run and audited.
Validation gates
-
In silico
Cross-validated on held-out cohorts with external benchmarks.
-
In vitro
Replicated in patient-derived, disease-relevant cellular systems.
-
In vivo / ex vivo
Efficacy and safety signals confirmed in whole-organism or tissue models.
-
Clinical cohort
Biomarker performance verified in prospectively defined patient groups.
Team & governance
Depth across the whole pipeline
Computational scientists, bench biologists and clinical developers work in one team, so no hand-off loses the biology.
-
Computational biology & ML
Model development, causal inference and data engineering for multi-omic and clinical data.
Multi-omicsCausal ML
-
Translational wet-lab science
Assay development and disease-model biology that turns predictions into testable results.
Assay designDisease models
-
Clinical development & regulatory
Trial design, biomarker strategy and regulatory planning from first-in-human onward.
Trial designRegulatory
-
Scientific & clinical advisors
Independent experts who review programs and challenge assumptions at every gate.
Independent reviewGovernance