Why Rare Disease Research Is Changing
For most of medical history, rare diseases have been an afterthought. With small patient populations and no obvious commercial market, pharmaceutical companies had little incentive to invest. Researchers struggled to find enough patients for meaningful studies. Doctors went entire careers without seeing a single case. The result was decades of diagnostic odysseys, recycled textbook paragraphs, and patients left to figure things out on their own.
That's shifting — and the reason matters. Rare diseases aren't just rare. They're windows into how the human body actually works. When a single gene mutation causes soft tissue to turn into bone, or prevents cells from recycling damaged proteins, or traps cholesterol inside cells that need it to flow freely — it teaches us something fundamental about systems that affect everyone.
Researchers sometimes call rare diseases "nature's knockout experiments." Each one isolates a single biological pathway and shows, with unusual clarity, what happens when it breaks. That knowledge doesn't just help the handful of people with the condition. It feeds directly into understanding cancer, neurodegeneration, osteoporosis, metabolic disease — conditions that affect hundreds of millions of people worldwide.
Three Diseases, Three Biological Systems
This page focuses on three conditions that illustrate the pattern. Each one is ultra-rare. Each one has a known genetic cause. And each one is teaching researchers something that extends far beyond the disease itself:
Fibrodysplasia Ossificans Progressiva (FOP) — a mutation in the ACVR1 gene causes the body to build bone where it shouldn't, progressively turning muscles, tendons, and ligaments into a second skeleton. Studying FOP has transformed our understanding of bone morphogenetic protein (BMP) signaling, a pathway that matters for fracture repair, heterotopic ossification after surgery, and bone cancer metastasis.
NGLY1 Deficiency — the loss of a single enzyme, N-glycanase 1, means cells cannot properly dismantle misfolded proteins. The resulting accumulation disrupts the nervous system, liver, and multiple other organs. NGLY1 research has revealed new details about ER-associated degradation (ERAD), a protein quality-control system implicated in Parkinson's disease, ALS, and age-related neurodegeneration.
Niemann-Pick Disease Type C (NPC) — mutations in NPC1 or NPC2 prevent cholesterol from exiting the lysosome, causing lipid buildup that progressively damages neurons. NPC research has expanded our understanding of intracellular lipid trafficking — a process relevant to Alzheimer's, atherosclerosis, and metabolic syndrome.
The rarer the condition, the cleaner the genetic signal. And a clean genetic signal is exactly what drug development needs. Studying conditions that affect dozens of people can generate insights that benefit millions.
What AI Can Do for Rare Conditions
AI doesn't replace doctors or researchers. What it does is make small datasets useful in ways that were previously impossible. For conditions with fewer than a few hundred known cases worldwide, that changes everything.
Find Patterns in Small Groups
AI can detect shared genetic and phenotypic traits even when patient numbers are tiny. For FOP, machine learning models have analyzed ACVR1 mutation variants against clinical outcomes to map which mutations produce the most aggressive bone formation — something impossible to see in a dataset of 800 patients without computational pattern recognition.
Speed Up Diagnosis
The average rare disease patient sees seven or more doctors over four to five years before receiving a correct diagnosis. AI tools like Face2Gene use facial recognition to match patient images against known genetic syndrome composites, and differential diagnosis systems like Isabel DDx generate condition shortlists from symptom descriptions. For NPC, where psychiatric symptoms often lead to years of misdiagnosis, AI-assisted symptom screening could dramatically shorten the diagnostic odyssey.
Suggest Existing Drugs That Might Help
AI platforms can scan vast libraries of approved medications to find potential matches for rare diseases. TxGNN, a graph-neural-network system developed at Harvard, analyzed treatment relationships across more than 17,000 conditions and identified drug-repurposing candidates including existing medications that had never been tested against ultra-rare targets.
These systems don't discover cures. They generate scientifically informed hypotheses by mapping connections between genes, diseases, and drugs. Researchers must still validate these leads through experiments and clinical studies, but AI dramatically accelerates the search — critical when patient populations are too small for traditional pharmaceutical investment.
Read and Organize Patient Experiences
Natural language processing turns unstructured patient stories, case reports, and forum posts into organized, searchable data. For NGLY1 deficiency — where fewer than 150 patients are known worldwide — AI has been used to extract symptom patterns from published case literature, identifying that alacrima (inability to produce tears) is a near-universal diagnostic marker that clinicians frequently overlook.
Predict How Conditions Progress
AI models use genetic and longitudinal patient data to forecast disease progression. For NPC, where the age of onset determines prognosis (early infantile, late infantile, juvenile, or adult), predictive models are helping researchers define natural history endpoints that clinical trials need before they can even begin. For FOP, AI analysis of flare-up patterns is starting to reveal which patients may progress faster — enabling earlier intervention.
For the tools, communities, and organizations where you can take action right now —
Browse ResourcesThree Systems, Three Failures
Each of these conditions isolates a different fundamental biological system. Understanding how that system breaks in a handful of patients teaches us how it works in everyone.
Fibrodysplasia Ossificans Progressiva
ACVR1What Breaks
The ACVR1 gene encodes a receptor in the bone morphogenetic protein (BMP) signaling pathway. In FOP, a gain-of-function mutation makes the receptor constitutively active. The normal pathway — injury triggers BMP signaling, which triggers bone repair — becomes permanently switched on. The body forms bone in muscles, tendons, ligaments, and fascia where none should exist.
Even mild trauma — a bump, a fall, an injection — can trigger a flare-up. The body responds to soft tissue inflammation by laying down new bone through endochondral ossification, the same process that builds a fetus's skeleton. Over time, a second skeleton forms around the first.
Key Clinical Features
Malformed big toes at birth are the most important early diagnostic clue — and one doctors frequently miss. Children develop episodic flare-ups: inflammatory swelling in muscles that eventually produces bone. Progressive joint fusion follows, typically beginning in the neck and shoulders and moving downward. Rib cage restriction causes thoracic insufficiency syndrome, one of the leading causes of death. Median life expectancy is around 40 years.
Why It Matters Beyond FOP
FOP research revealed that activin A — previously thought harmless to bone — incorrectly triggers ossification through the mutant receptor. This discovery reshaped understanding of heterotopic ossification, which also occurs after traumatic injuries, joint replacements, and blast injuries in combat veterans. The same BMP pathway is implicated in bone metastasis in cancer, making FOP research directly relevant to oncology.
NGLY1 Deficiency
NGLY1What Breaks
Cells constantly produce proteins that fold incorrectly. A quality-control system called ER-associated degradation (ERAD) catches these misfolded proteins, strips their sugar chains, and sends them to the proteasome for destruction. The enzyme N-glycanase 1 (NGLY1) handles the sugar-stripping step. Without it, damaged glycoproteins accumulate inside cells, triggering cellular stress that is particularly destructive to neurons.
Key Clinical Features
Most patients show symptoms in infancy: severe developmental delay, movement disorders, low muscle tone, liver dysfunction, and seizures. The hallmark diagnostic clue is alacrima — the inability to produce tears when crying. It's present in nearly every known patient and is frequently the detail that leads a clinician to suspect NGLY1 deficiency, yet it often goes unnoticed for years.
Why It Matters Beyond NGLY1
NGLY1 research cracked open the ERAD pathway in ways that have implications for Parkinson's disease, ALS, and age-related neurodegeneration — all conditions linked to misfolded protein accumulation. The same protein recycling machinery that fails in NGLY1 deficiency is implicated in the toxic aggregates that define Alzheimer's and Huntington's disease. Understanding precisely where and how that machinery breaks down in NGLY1 patients gives researchers a cleaner model than any animal study could provide.
A Model for Patient-Driven Research
NGLY1 deficiency is also famous as a landmark in patient-driven research acceleration. Parents of the first diagnosed child launched the Grace Science Foundation, funding multiple labs simultaneously, building open patient registries, and sharing data openly. The model — patients funding and coordinating research rather than waiting for institutional interest — has become a template for other ultra-rare communities.
Niemann-Pick Disease Type C
NPC1 · NPC2What Breaks
Cholesterol doesn't float freely through cells. It has to be actively transported from one compartment to another. The NPC1 and NPC2 proteins manage cholesterol's exit from the lysosome — the cell's recycling centre. When either protein is missing or dysfunctional, unesterified cholesterol and other lipids accumulate inside lysosomes, eventually overwhelming the cell. Neurons are particularly vulnerable. NPC1 mutations account for roughly 95% of cases.
Key Clinical Features
NPC presents differently depending on when it begins. Early-onset patients show enlarged liver and spleen and neonatal jaundice. As the neurological component develops, patients experience progressive ataxia, speech difficulties, seizures, cognitive decline, and psychiatric symptoms. The most characteristic clinical sign is vertical supranuclear gaze palsy — the inability to move the eyes vertically. Age of onset determines severity: early infantile (under 2 years) progresses fastest, while adult-onset forms can remain undiagnosed for decades, often mischaracterized as psychiatric illness.
Why It Matters Beyond NPC
NPC research has fundamentally expanded our understanding of intracellular cholesterol trafficking. The same lysosomal transport system that fails in NPC is implicated in Alzheimer's disease — NPC1 variants are now recognized as risk factors for late-onset Alzheimer's. The lipid-storage mechanism also connects to atherosclerosis, metabolic syndrome, and age-related macular degeneration. Research into how NPC drugs mobilize trapped cholesterol is informing therapeutic strategies across all of these conditions.
What's Been Tried — and What It Means
None of these three conditions has a cure. Treatment is either off-label, experimental, or focused on slowing progression rather than reversing it. But the therapeutic landscape for each reveals how rare disease research generates leads that extend far beyond the original condition.
Palovarotene
FOP · Retinoid receptor agonist
The first drug specifically developed for FOP. Palovarotene inhibits new bone formation by blocking the BMP signaling pathway downstream of the ACVR1 mutation. It received regulatory approval in Canada in 2023, making it one of the first approved therapies for any ultra-rare skeletal condition. It slows new heterotopic ossification but does not reverse existing bone.
Approved in Canada · Under review elsewhereACVR1 Inhibitors
FOP · Targeted receptor blockade
Multiple companies are developing antibodies and small molecules that block the ACVR1 receptor directly. These aim to shut down the overactive signaling at its source. The challenge is specificity — ACVR1 has normal roles elsewhere in the body, so blocking it completely may cause off-target effects.
Clinical trials ongoingMiglustat
NPC · Substrate reduction therapy
Originally developed for Gaucher disease, miglustat reduces the production of glycosphingolipids — one of the lipid types that accumulate in NPC. Clinical use has shown stabilization or slowing of neurological decline in some patients, though it does not address the underlying cholesterol transport defect.
Approved in EU · Off-label elsewhereCyclodextrin (HPβCD)
NPC · Cholesterol mobilization
Hydroxypropyl-beta-cyclodextrin can extract trapped cholesterol from lysosomes, directly addressing the core NPC defect. It has shown efficacy in animal models and compassionate use cases. Phase 2/3 clinical trials are ongoing. The challenge is delivery — the compound doesn't cross the blood-brain barrier easily, requiring intrathecal (spinal) administration for neurological benefit.
Most promising — trials ongoingGene Therapy Approaches
NPC + NGLY1 · Viral vector delivery
Both NPC and NGLY1 deficiency are targets for AAV-based gene therapy — delivering a functional copy of the missing or mutated gene directly to affected cells. For NPC, gene therapy targeting NPC1 has shown dramatic lifespan extension in mouse models. For NGLY1, the Grace Science Foundation is funding gene therapy research as one of its primary pipeline candidates.
Preclinical · multiple programsPatient-Driven Drug Discovery
NGLY1 · Grace Science Foundation model
The Grace Science Foundation funds multiple simultaneous research programs, from enzyme replacement concepts to small-molecule activators of alternative protein-processing pathways. Their model — patients directing and funding research, not waiting for institutional interest — has produced more therapeutic leads in a decade than many rare conditions see in a century.
Model for ultra-rare researchThe Pattern Across the Ultra-Rare
Hundreds of conditions share the same structural reality: a known genetic cause, a handful of published case reports, and almost nothing else. The gaps below aren't unique to FOP, NGLY1, or NPC — they're a pattern that repeats across every ultra-rare disease, including Hajdu-Cheney Syndrome.
No patient registries. Most ultra-rare conditions have no structured registry on any major platform. Even a small patient-reported outcomes registry would transform the evidence base overnight — but someone has to build it.
No natural history studies. Without longitudinal tracking of how a disease progresses over time, clinical trials have no endpoints to measure against. Researchers can't prove a drug works if they don't know what "untreated progression" looks like.
No clinical trials. FOP achieved its first approved drug only after decades of preclinical work and patient advocacy. Most ultra-rare conditions haven't even reached the preclinical stage. The pipeline from lab bench to patient bedside barely exists.
Diagnostic delays measured in years. The average rare disease diagnosis takes four to five years. For conditions like NPC, where psychiatric symptoms dominate the early presentation, patients can go decades without a correct diagnosis. AI diagnostic tools are starting to close this gap, but adoption is slow.
No accessible translation of the science. When research does exist, it's locked behind paywalls and written for specialist audiences. Patients and families — the people most affected — often can't access or understand the literature about their own conditions.
Fragmented patient communities. With fewer than a few hundred patients worldwide, most ultra-rare communities have no central meeting point. Patients don't know each other. Researchers don't know where to find them. The infrastructure that would connect the two doesn't exist.
The pattern is the same everywhere: a registry, a natural history study, and a patient-accessible translation of the science — those three things alone would represent transformative progress for any one of these conditions. That's what Bare Your Rare is trying to build.
If you're a researcher, clinician, or patient interested in contributing to rare disease research infrastructure — whether for FOP, NGLY1 deficiency, NPC, Hajdu-Cheney Syndrome, or any other ultra-rare condition — we want to hear from you. The biggest barrier isn't the science. It's the lack of connected people.
Every rare condition is a natural experiment.
The more we study them, the more we learn about conditions affecting everyone. Patient registries give structure. AI finds patterns. Research institutions provide validation. For the first time, patients are at the centre of it.