Organ models · Drug discovery
Organoid Biofoundry
A publication about growing human tissue that behaves enough like an organ to answer a question. We cover the models, the assays built on them, and the reproducibility problems that decide whether any of it reaches a clinic.
Latest analysis
October 26, 2026 Digits from a dish: organoids that grow fingers obey fluid physicsMouse limb-mesenchyme organoids treated with Fgf8b and Wnt3a spontaneously break symmetry and elongate digit-like protrusions. Coarse-graining the validated cell-based model yields a Cahn-Hilliard equation whose interface instability is mathematically the fingering phenomenon of fluid physics, suggesting organoid shape is a controllable physical parameter rather than decorative noise.
October 26, 2026
The cornea's own matrix proteins sound the antiviral alarm, and organoids hear only half of it
Mice lacking the stromal proteoglycans lumican or biglycan develop worse HSV-1 keratitis, clear virus poorly, and recruit few neutrophils, monocytes or plasmacytoid dendritic cells early. The cornea answers infection by switching on these matrix proteins in its epithelium to amplify CCL2, a human iPSC corneal organoid recapitulates the epithelial half of that circuit, and the missing half defines exactly what the model can and cannot certify for drug screens.
October 26, 2026
The healthy-baseline problem: a UK-trained brain reference works in China, but only on the model's terms
Stochastic cortical self-reconstruction builds a personalized healthy reference for brain structure and flags atrophy as deviations from it. Transferred from UK Biobank to an independent Chinese cohort, it keeps working with no adaptation, peaks after fine-tuning (average AUC 0.848), and exposes a capacity trap: the 20-million-parameter model overfits a 640-subject local cohort while the smaller structured model does not. The same reference-frame arithmetic governs every organoid quality-control baseline and digital endpoint in drug development.
October 24, 2026
Bladder organoids get a pre-surgical decision role
The POLO trial in Switzerland plans to screen patient-derived organoids from low-grade bladder tumors against four intravesical chemotherapies and treat patients with the predicted best agent before surgery. It is a 28-patient, single-arm phase II, and its design exposes exactly what an organoid assay must prove to earn a place ahead of the operating room.
October 24, 2026
Organoid-guided salvage therapy, uncontrolled
A Tianjin trial plans to establish organoids from 40 patients with refractory breast cancer, screen them against marketed drugs, and treat each patient with the predicted most sensitive agent. The design shows how far organoid-guided therapy has spread clinically, and how thin the evidence under it still is.
October 24, 2026
A hidden chromatin test for organoid fidelity
A bioRxiv preprint uses CRISPR to delete a neural enhancer sitting 550 kb from SOX2, then watches the phenotype surface only in neural stem cells, teratomas, and cerebral organoids. The result doubles as a stress test for how organoid quality is actually measured.
October 23, 2026
A ferroptosis drug claim dismantled by two knockouts and an organoid
MMRi62 was reported to kill cancer cells by degrading ferritin and triggering ferroptosis. CRISPR knockouts of ferritin heavy chain and of NCOA4 in medulloblastoma cells, plus c-Myc/OTX2 tumors grown in cerebral organoids, show the compound actually drives iron-depletion apoptosis as a direct iron chelator.
Start here
Standing explainers that do not go stale. Read the primer first if this field is new to you.
The primer
What is an organoid?
The biology behind the models: how a stem-cell-derived 3D tissue is grown, quality controlled, and read out, and where the resemblance to a real organ stops.
Model
Brain organoids
Self-organized cortical-like tissue: what develops, what does not, and why maturation is the binding constraint.
Model
Liver organoids
Metabolic and toxicological readouts, and how far they predict human hepatotoxicity.
Model
Kidney organoids
Nephron-like structures, off-target cell types, and what that means for injury models.
Constraint
The vascularization problem
Why organ models stop growing at a millimetre, and the approaches trying to get past it.
Sections
Section
Organ models
Each model reproduces some of an organ and omits the rest. These pages set out what is faithfully modelled, what is missing, and which questions the model can therefore be trusted to answer.
Section
Drug discovery
Patient-derived models, tumour screening, and high-throughput assays that have to survive a real pipeline.
Section
Analysis
Daily readings of new research, each asking what the work changes for this field.
How we work
We report what a model was validated against, not what it was advertised to do. Sample sizes, donor counts and passage numbers matter for reproducibility, so we quote them where the source does and note their absence where it does not. Preclinical results are reported as preclinical: an effect in a dish is not an effect in a patient, and we do not blur that line to make a headline. The full method, including how pieces are selected and produced, is on the about page.