Drug discovery · Autoimmune disease
Ditto Biosciences
A three person company mining the genomes of viruses, ticks and worms for proteins that already know how to calm the human immune system. Preclinical, with no named drug candidate and about $500,000 on the public record.
The project that got put on ice
On May 14, 2025, Arcadia Science published a methods paper with an unusual epilogue. It described a pipeline for spotting viral proteins that imitate human ones, benchmarked against viral proteins known to mimic 11 human targets. For every one of the 11, the pipeline recovered at least one known mimic.
Then the authors closed it down. “We’ve put this effort on ice,” the page reads, under a tag that says strategic misalignment.
Two of the listed authors were Adair L Borges and Emily C P Weiss. Within months they had built a company around the idea Arcadia had just shelved, with a third Arcadia colleague, Dennis Sun. The Ditto Bio organization on GitHub holds two repositories. One is a fork of Arcadia-Science/2024-mimic-benchmarking, the code behind that paper.
The problem they picked
Autoimmune disease is a large market served by blunt instruments. The company puts it this way: “Autoimmune diseases affect roughly one in ten people worldwide, yet most treatments suppress immune pathways rather than restore immune tolerance.”
Ditto’s wager is about where new molecules come from. Its launch post argues that 90 percent of drugs fail the first time they meet a human, after three to five years and $15 million to $20 million spent in cells and mice. Its answer is to skip the invention step. Viruses hide from immune detection, ticks suppress inflammation at the bite, and intestinal worms hold down a host response for decades. Those proteins were optimized in the only system that matters, and most have never been looked at: the company says more than 98 percent of viral and parasite proteins remain uncharacterized.
This is not a new thought, and the company does not claim it is. What changed is cheap sequencing plus structure prediction. AlphaFold made it possible to ask, at proteome scale, which parasite protein grabs which human one.
Three scientists who had already worked together
Dennis Sun is chief executive. He holds an AB from Harvard and a PhD from UC Berkeley, and was chief of staff at Arcadia Science, where he built Protein Cartography, an open source pipeline that uses AlphaFold structures to map protein similarity across the tree of life.
Adair Borges is chief scientific officer. Her PhD is from UCSF, in anti CRISPR biology, and she was afterwards a Miller Fellow at UC Berkeley. PubMed indexes 17 of her CRISPR papers between 2017 and 2025, in Science, Cell, Nature Microbiology and mBio.
Emily Weiss is chief technology officer, a UCSD PhD who was at Illumina and DuPont before Arcadia. She is the founder who pitched at DeepMind’s startup sessions at Google in 2026.
They had run together for three years before starting anything. As one of them put it in a company post: “Rewind to 2023, the three of us worked together at Arcadia Science.”
The first result they showed
On March 2, 2026 Ditto published its own analysis, the most testable thing it has put in public. Using a platform it calls MoleculeMapper, the team mapped roughly 10,000 proteins from human infecting viruses onto their human targets, then asked a narrow question: if you rank candidate drug targets for rheumatoid arthritis and add one feature, how often a virus attacks that target, do the targets that already have approved drugs move up?
They did. Clinically approved targets rose by an average of 38 percent in the ranking. Tumor necrosis factor, the target behind the first approved biologics, moved from 82nd to 12th. The company also notes that etanercept, one of those blockbusters, uses the same mechanism that viral TNF inhibitors arrived at on their own.
It is a retrospective check on known answers, not a drug. But it is a real number attached to a real method.
What is proven, and what is still claimed
| Evidence | What the record shows | Source type |
|---|---|---|
| FDA and trials | No 510(k), no PMA, no registered clinical trial as of Sep 24, 2026. No investigational new drug application, no named lead molecule, no disclosed lead indication. | Not found |
| Papers under the company | No PubMed record for Ditto Bio or Ditto Biosciences as an affiliation, searched Sep 24, 2026. | Not found |
| Founder publication record | Adair Borges appears on 17 PubMed indexed CRISPR papers from 2017 to 2025, including Science (Oct 2018), Cell (Aug 2018 and Sep 2019), Nature Microbiology (May 2020 and Jan 2022) and mBio (Sep 2025). | Public record |
| Prior art from Arcadia | “A method for computational discovery of viral structural mimics,” Arcadia Science, May 14, 2025. Authors including Adair L Borges and Emily C P Weiss. Foldseek plus Gaussian mixture modeling, benchmarked on viral proteins known to mimic 11 human proteins, at least one recovered for each. | Public record |
| Patents | No patent found assigned to Ditto Bio or Ditto Biosciences. An inventor search returns EP3976797A2, “Anti-crispr inhibitors,” assigned to the Regents of the University of California, published Apr 6, 2022, EP and US family members active. | Public record |
| Platform result | Company blog, Mar 2, 2026: about 10,000 viral proteins mapped to human targets; rescoring the top 100 rheumatoid arthritis targets by viral targeting frequency lifted clinically approved targets by an average of 38 percent; TNF moved from rank 82 to 12. Not peer reviewed. | Company-stated |
| Scale and binding | Launch post, Feb 2026: more than 1 million proteins analyzed across primate infecting viruses, ticks and parasitic worms; thousands predicted to hit clinically validated human targets; binding as tight as 1 to 2 nM. No underlying data published. | Company-stated |
| Non dilutive funding | No NIH or NSF award found to the company. | Not found |
| Securities filings | No SEC filing under either name in EDGAR company search or full text search as of Sep 24, 2026. | Not found |
| Where it is | Y Combinator, Tracxn and Preqin list San Francisco. PitchBook lists a corporate office at 815 Solano Avenue, Albany, CA. | Sources differ |
Read plainly: a research company fourteen months old, with one published analysis, a founding team whose track record checks out, and nothing yet that a regulator or a journal has seen. The binding and scale figures are the company’s own and cannot be verified from outside. The distance to a first human dose is measured in years and in money it has not raised.
What to watch
- A named lead molecule and an indication. Everything so far is platform and target selection.
- A preprint or a second post carrying the wet lab data behind the 1 to 2 nM binding claim.
- The tissue biobank the company says it is building to predict immunogenicity, the hardest problem for novel protein therapeutics.
- A pharma partnership. The launch post asks industry immunologists and biologics developers to get in touch, which is a partnering strategy.
- A priced round with a Form D behind it. Nothing has been filed.
In their words
“We believe the world’s best immunologists live inside of you.”
Ditto Bio founders, Launch YC post, February 2026 · Company
“We’ve decided not to pursue this project because it doesn’t play to the unique strengths of our platform. We’re sharing it in the hope that it will enable future research in this area outside of Arcadia.”
Arcadia Science, mimicry methods paper, May 2025 · Prior employer
“Rewind to 2023, the three of us worked together at Arcadia Science.”
Ditto Bio co-founder, LinkedIn, 2026 · Company
“Autoimmune diseases affect roughly one in ten people worldwide, yet most treatments suppress immune pathways rather than restore immune tolerance.”
Ditto Bio, LinkedIn, 2026 · Company
“If nature’s best drug discovery engine is evolution, and if parasites have spent hundreds of millions of years evolving proteins specifically designed to control the human immune system, then those proteins are essentially ready-made drug candidates sitting in an unread library.”
StartupHub.ai, July 2026 · Independent
“While not necessarily an entirely novel approach, the past decade has yielded significantly reduced costs of sequencing leading to a proliferation of genomic and transcriptomic data not just around human health and disease but across organisms.”
Decoding Bio, BioByte 150, March 2026 · Independent
“This is pre-product, pre-candidate, pre-revenue biotech at the earliest possible stage; the YC model fits awkwardly here as drug discovery requires 7-10 years and $100M+ to reach meaningful milestones.”
The YC Tier List, 2026 · Independent
“Ditto is taking a fundamentally different approach by learning from nature’s own experiments and translating them into scalable, engineered solutions.”
Multimodal Ventures, portfolio announcement, March 2026 · Investor
Related companies
Sources
- Public recordSEC EDGAR company and full text search, no results
- Public recordPubMed searches, company affiliation and Borges AL
- Public recordGoogle Patents, EP3976797A2 Anti-crispr inhibitors
- Public recordA method for computational discovery of viral structural mimics
- Public recordDitto Bio GitHub organization
- IndependentDitto Bio Review: AI Parasite Drug Discovery
- IndependentBioByte 150
- IndependentTop Bio and Health Startups in YC W26
- IndependentDitto Biosciences review
- IndependentCompany profile, funding and headcount
- IndependentCompany profile, deals and investors
- IndependentDitto Biosciences asset profile
- IndependentPioneer Fund therapeutics portfolio listing
- IndependentFlex Capital Q2 2026 highlights
- InvestorDitto Bio joins the Multimodal Ventures portfolio
- CompanyLaunch YC: Ditto Bio, evolutionary therapies for autoimmune disease
- CompanyMapping parasite molecules to treat autoimmune disease
- CompanyCompany and jobs pages, founder bios
- CompanyDitto Bio website, founding team and approach
- CompanyLinkedIn page: advisors, FAST cohort, DeepMind pitch, Arcadia origin
Profile researched and written by Healthcare Discovery. Last updated September 29, 2026.
