Probably Genetic, an AI platform powering the research, diagnosis and treatment of genetic diseases, has raised $12 million in series A funding, Fierce Healthcare has learned.
The round was led by Threshold, with participation from Y Combinator and other existing investors. The capital brings the company’s total to $14 million and will be used to grow the core part of the business, for hiring and to enter the drug discovery arena.
The company’s primary product is a physician-ordered genetic test for undiagnosed rare disease patients, who typically face a seven-year diagnostic odyssey before getting answers. There are two pathways to testing: free testing, through pharma partnerships, and insurance-billed testing. The free programs address specific conditions, with a limited number of tests available at a time.
Patients can check their eligibility by completing a health and diagnostic assessment online. Probably Genetic’s AI platform converts self-reported symptoms, medical records and biological data into phenotypic data that it hopes can be used to train AI models for precision medicine in the future. Thus far, it has collected data from 130,000 patients.
“The bottleneck in drug development isn’t designing molecules anymore. It’s knowing which targets are worth pursuing. Drugs backed by human genetic evidence are roughly twice as likely to succeed, yet data linking genes to real clinical outcomes over time barely exists,” Emily Melton, managing partner and co-founder at Threshold Ventures, told Fierce Healthcare in an emailed comment. “Probably Genetic's AI platform has built that dataset as a byproduct of helping genetic disease patients get diagnosed. … They’re ready to put their data to work with biopharma partners to find the next generation of drug targets.”
While today, Probably Genetic collects saliva samples for genetic testing, it plans to expand to proteomic testing in 2027, which requires blood draws. Proteomic sequencing data is crucial for drug discovery. This testing is not something patients would pay for, nor is it something that would be reimbursed through insurance. Instead, this will be paid for by pharma partners.
In August, Probably Genetic was awarded up to $10 million from ARPA-H, an agency within the Department of Health & Human Services that funds high-risk, high-reward research. The company is participating in the agency’s new Rare Disease AI/ML for Precision Integrated Diagnostics (RAPID) program, leveraging its dataset to identify undiagnosed patients and give drug developers real-world data needed for drug discovery and clinical trials.
“We can already train AI models that can cut that diagnostic delay short, but to train those models, you need really good data—and the data to train these models doesn’t exist today,” Lukas Lange, Ph.D., co-founder and CEO, told Fierce Healthcare in an exclusive interview.
When Lange was a genetics doctorate student at Oxford, he was digging through thousands of documents of patient data to identify them as rare disease patients. But this information was highly fragmented and often lacking observable traits, like physical characteristics or behaviors. That’s where the idea for Probably Genetic was born—to gather all the data in one place from the patient and eventually help identify others with similar symptom profiles.
The vast majority of patients going through Probably Genetic don’t pay for the testing, as it is covered by both commercial and government plans, per Lange. The remaining few either have copays, typically capped at $250, or get the testing covered through the company’s pharma partnerships. Pharma’s incentive is to identify patients who can participate in clinical trials or are a fit for therapies already on the market. The company works with UCB, Biogen, Ultragenyx, Chiesi, Bial, Mahzi Therapeutics, X4 Pharmaceuticals and undisclosed others.
Probably Genetic does not have its own labs, working with half a dozen partners today to run the sequencing. It doesn’t see these other labs as its competition. “The biggest game here is actually not played on the diagnostic side, because unfortunately there is such a vast pool of patients out there that needs to get diagnosed,” Lange said. “The competition is going to be in actual discovery.”
Most genetic diseases have no treatments. That’s why Probably Genetic hopes to enter the drug discovery arena. The ideal scenario, laid out by Lange, would be getting someone personalized preventive therapeutics based on their genetic testing, before they even experience symptoms of their illness.
“Now we’re finally seeing this era where the combination of increasing availability of data and algorithms and compute actually unlocks a future where personalized medicine is possible,” Lange said.
Probably Genetic was founded in 2018 and is profitable. It claims to help diagnose over half of patients who previously had unexplained symptoms.