Diagnostics · Autoimmune treatment selection
Coral Genomics
A San Francisco startup that dosed a patient’s own immune cells in a dish with the drugs their doctor might prescribe, then read out which one the cells answered to. It enrolled 200 patients, certified a lab, and stopped. The trial registry, not a press release, records how it ended.
Nine words on a government form
Most company obituaries get written by reporters. This one was filed on a federal form.
On May 18, 2023, someone at Coral Genomics moved study NCT04995783 from active to terminated on ClinicalTrials.gov. The form asks why a study stopped. The answer entered was nine words: “Coral underwent an APA and is no longer in operation.”
APA is an asset purchase agreement. The study, called REEF, had reached its actual completion date of August 1, 2022, with 200 patients enrolled across ulcerative colitis, Crohn’s disease, rheumatoid arthritis and psoriasis. The results have never appeared in a journal.
Darts at a wall
Autoimmune medicine has a selection problem oncology largely solved. A patient with rheumatoid arthritis or Crohn’s disease gets offered a biologic costing tens of thousands of dollars a year that works in roughly half the people who take it. There is no routine test for which half.
Jonathan Rosenberg, a gastroenterologist at Illinois Gastroenterology Group who referred patients into Coral’s study, told GenomeWeb in 2021 that clinicians have the drugs but no tool to choose among them. For patients nothing works on, the journey can end in bowel surgery, and everything before it is thrown “like darts at a wall in the hope something sticks.”
The money is why people keep trying. Coral said in January 2022 that US autoimmune medication spending runs over $70 billion a year and has passed cancer drugs. A year earlier it used $54 billion. Both were company-stated.
Two students from the Broad
Atray Dixit and Jacob Borrajo met at the Broad Institute in 2014, entered an NIH crowdsourced proposal challenge on a whim, and won first place. Dixit did his doctorate in Aviv Regev’s lab and was first author in 2016 on Perturb-seq, which let CRISPR edits be screened with single-cell RNA sequencing. Borrajo had already founded Scribio. They got into Y Combinator with a sequencing cost idea, and Dixit started the batch on the day of his MIT graduation ceremony.
The motive was closer to home than the technology suggests. “Initially, I became motivated to solve this problem by having close family members develop autoimmune conditions and witness their challenges.”
What it actually built
Coral’s first asset was cost. Its sample pooling method cut sequencing costs by about half, and the grant abstract describes “a new sample pooling approach capable of sequencing hundreds of samples with less than one week of wait time” plus a “real-time analysis platform capable of returning variant calls within two hours of sequencing completion.” It was never sold as a product. It was what made the real work affordable.
The real work was functional. Coral took primary immune cells from enrolled patients, exposed them to different drug classes, and read the response by RNA sequencing, often at single-cell resolution. That became the Functional Response Score. In October 2020 GenomeWeb reported the lab had “recently received CLIA certification.” In September 2021 Dixit wrote that “we have completed the CLIA certification process for an NGS-based diagnostic test.”
The pitch was coverage as much as accuracy. Coral drew on UK Biobank and BioBank Japan, paid DrugViu per lead to recruit Black and other minority patients, and wanted to “cross-validate our discovery of a novel subpopulation of patients that don’t appear to respond to any existing therapies on the market in this cohort.” True non responders, if real, are a drug target rather than a test result.
Two good years, then nothing
2020 and 2021 were the peak: a LEO Pharma pilot, a $2 million NHGRI Phase II award in October 2020, the DrugViu deal in March 2021, the Lyfebulb and Arena IBD Innovation Challenge and its $25,000 prize that September, and in January 2022 an agreement with the Crohn’s & Colitis Foundation to test the score in up to 1,000 IBD Plexus patients.
Then the record thins. The last NIH obligation is fiscal 2021. REEF completed in August 2022. The registry entry changed in May 2023. Asked in 2021 to name the biggest hurdle ahead, Dixit had answered: “Publishing the results from our independently run clinical studies.” It was never cleared.
The founder documents the ending himself. On LinkedIn he writes that “After a four year journey towards creating precision medicine diagnostics, we were acquihired by Gordian Biotechnology.” He became VP of Technology there, left, and now runs Oncko, a cancer combination trial company. Coral’s LinkedIn page survives, renamed Coral (acquihired). Both domains are gone: coralgx.com was parked on Sedo by September 2024, and coralgenomics.com now serves an unrelated gambling site.
What is proven, and what is still claimed
| Evidence | What the record shows | Source type |
|---|---|---|
| REEF study | NCT04995783, observational case control. Started Sep 1, 2020. Enrollment 200, actual. Primary completion Aug 1, 2022. Four conditions, one site. Collaborator NHGRI. Terminated May 18, 2023. | Public record |
| Study as marketed | Company materials described a larger program covering up to 1,000 IBD Plexus patients. Registered enrollment stopped at 200. | Differs from registry |
| Peer-reviewed validation | None found for the Functional Response Score as of Sep 23, 2026. PubMed carries a 2021 adverse event deep learning paper with Dixit as senior author. | Not found |
| CLIA laboratory | Reported by GenomeWeb in Oct 2020, repeated by Dixit in Sep 2021. No CMS certificate number found. | Independent |
| Federal grants | R43HG010445 $142,674 (FY2019); R44HG010445 $794,628 (FY2020) and $1,205,206 (FY2021), NHGRI; R44ES032515 $811,373 (FY2021), NIEHS. Total $2,953,881. | Public record |
| Grant totals as described | GenomeWeb described two 2020 grants “totaling approximately $3.6 million.” Dixit said NIH grants were double a $2 million seed, implying about $4 million. RePORTER shows $2.95 million. | Claim above the record |
| Patent | CB Insights lists one application filed Dec 15, 2020, status Application. Dixit in 2021: “Part of it is trade secret and some of it is covered by a patent that has not yet been issued.” No granted patent found. | No grant found |
| Partnerships | LEO Pharma (2020), DrugViu (Mar 2021), GI Alliance and an unnamed hospital network (2021), Crohn’s & Colitis Foundation (Jan 2022). No results reported from any. | Company-stated |
| Operating status | Registry says no longer in operation, May 2023. LinkedIn page renamed Coral (acquihired). YC record lists Acquired. Both domains lapsed. | Public record |
Read plainly: Coral built a real lab capability, won real federal money, and enrolled real patients in four diseases at once. It never produced what would have made any of it purchasable. No published validation, no payer decision, no named customer, no FDA record. Its central claim was tested in a 200 patient study whose results were never released.
What to watch
- Whether the REEF data ever appears. Two hundred patients with paired functional assays is worth publishing even now.
- The patent application filed December 15, 2020. No issued patent has been found under the company name.
- What Gordian Biotechnology did with the acquired assets. Gordian works on in vivo screening for age related disease.
- The non responder subpopulation Coral claimed to have found. Nobody has reproduced it in public.
- Scipher Medicine’s PrismRA as the benchmark: published validation, then a Medicare local coverage determination in September 2023. That is the sequence Coral did not finish.
In their words
“Coral underwent an APA and is no longer in operation.”
Reason for stopping, ClinicalTrials.gov NCT04995783, entered May 18, 2023 · Public record
“Right now, we have a wide array of medications to try to target various inflammation pathways to address underlying causes, but we don’t have any tools to decide which medication is right for the patients.”
Jonathan Rosenberg, gastroenterologist, Illinois Gastroenterology Group, GenomeWeb, 2021 · Independent
“Sample sizes for disease risk and drug response studies are small and often phenotype is quite hard to define.”
Atray Dixit, co-founder and CEO, GenomeWeb, May 2021 · Independent
“Patients and doctors often face a painful trial-and-error process to find a medication that works for them.”
Atray Dixit, partnership statement, January 2022 · Company statement
“While advances in IBD treatments have enabled improvement in clinical outcomes, development of tools that can predict which therapy will work best for each individual patient is critical.”
Angela Dobes, VP of IBD Plexus, Crohn’s & Colitis Foundation, January 2022 · Partner statement
“Publishing the results from our independently run clinical studies.”
Atray Dixit, asked to name the biggest hurdle ahead, BIOS Community, September 2021 · Interview
“Finally, on the funding side, after a $2M seed round post-YC we brought in double that amount in the NIH grants.”
Atray Dixit, BIOS Community, September 2021 · Above the record
“After a four year journey towards creating precision medicine diagnostics, we were acquihired by Gordian Biotechnology.”
Atray Dixit, LinkedIn post · Founder-stated
Related companies
Sources
- Public recordClinicalTrials.gov NCT04995783 (REEF)
- Public recordNIH RePORTER R43HG010445 and R44HG010445
- Public recordNIH RePORTER R44ES032515
- Public recordSEC EDGAR Form D search, no results for Coral Genomics
- IndependentCoral Genomics Bets on Diverse Data to Advance Precision Medicine for Autoimmune Disease
- IndependentStartup Coral Genomics Wins $2M SBIR Grant From NIH
- IndependentCoral Genomics, Crohn’s & Colitis Foundation Partner on Study of IBD Treatment Response
- InterviewFounder Spotlight #24: Atray Dixit @ Coral Genomics
- IndependentLyfebulb and Arena Pharmaceuticals announce the 2021 IBD Innovation Challenge winner
- IndependentCoral Genomics profile, patents and investors
- IndependentMedicare to cover PrismRA diagnostic test for rheumatoid arthritis
- CompanyCoral and the Crohn’s & Colitis Foundation announce an IBD partnership
- CompanyCoral Partners with DrugViu on Minority Representation in Autoimmune Precision Medicine
- CompanyCoral Genomics page, renamed Coral (acquihired)
- FounderAtray Dixit profile and post on leaving Gordian Biotechnology
- Domaincoralgx.com, Sedo domain parking page
- Domaincoralgenomics.com, now an unrelated site
Profile researched and written by Healthcare Discovery. Last updated September 29, 2026.
