Somewhere in a San Francisco laboratory, a database is growing. It contains something north of a million protein sequences—pulled from ticks, intestinal worms, viruses. Organisms that, by most measures, we'd prefer to avoid. Ditto Biosciences, barely six months old and fresh from Y Combinator's Winter 2026 cohort, is convinced these evolutionary survivors hold answers to some of medicine's most vexing problems.
The logic, when you hear it, has a certain elegance. Parasites have been perfecting the art of immune manipulation for millions of years. They feed on us, live inside us, reproduce within us—all while our immune systems remain, if not fooled, at least sufficiently pacified. A tick latches onto skin and drinks blood for days without triggering the inflammation that should, by all rights, alert the host and end the meal. Helminths colonize human intestines, dampening the very allergic and autoimmune responses that evolved to expel them.
What if those molecular tricks, refined across eons of host-parasite warfare, could be repurposed? Not the organisms themselves—experiments with live parasites as therapy have proven messier than hoped—but the individual proteins they deploy. The immune-quieting molecules. The inflammation dampers. The tools that let something foreign become, temporarily, tolerated.
It's a premise that sounds almost too neat. Nature as pharmacy. Millions of years of R&D, pre-completed. And yet precedent exists, scattered across pharmacology's history. Aspirin from willow bark, sure. But also exenatide from Gila monster venom, now a diabetes drug with billions in sales. Captopril, the first ACE inhibitor, inspired by compounds in pit viper venom. Ziconotide, a painkiller derived from cone snail toxin that's 1,000 times more potent than morphine.
The difference now, according to Ditto's founders, is scale. What AI and structural biology tools have made possible isn't finding a candidate from nature. It's screening a million of them, identifying thousands with predicted activity, and narrowing to the handful worth betting on. Whether that changes the math—whether parasitology becomes a viable pipeline rather than an academic curiosity—is the $72 billion question.
Money, Politics, and a Market Under Pressure
That figure isn't hypothetical. The autoimmune therapeutics market, analysts estimate, hit $72.1 billion in 2024 and should crest $92 billion by 2030. But the money is getting messier.
Medicare's drug price negotiation program, the most significant government intervention in pharmaceutical pricing in decades, kicked off in January. Ten drugs are in the first round. Two are heavy-hitters in autoimmunology: Enbrel (etanercept) and Stelara (ustekinumab). The Congressional Budget Office projects $6 billion in savings for 2026 alone. Pharma executives have used words like "catastrophic." Patient advocates counter that list prices had become unsustainable.
Simultaneously, biosimilars are eating into what were once fortress brands. Humira, AbbVie's juggernaut, lost U.S. patent exclusivity in 2023. Within a year, CVS Caremark shifted its formulary to favor Hyrimoz, one of several adalimumab biosimilars. The result: biosimilar share jumped from 2% to 19% overall, capturing 43% of new prescriptions by the end of 2024.
AbbVie, to its credit, saw this coming. The company spent years building its next generation—Skyrizi (risankizumab, an IL-23 inhibitor) and Rinvoq (upadacitinib, a JAK inhibitor). Combined sales of the two should exceed $31 billion by 2027, the company forecasts. It's a textbook case of planned obsolescence, except you're obsoleting your own multibillion-dollar asset.
Other players are pivoting too. Eli Lilly and Repertoire Immunotherapies inked a $2 billion autoimmune alliance in January 2026. Nektar and Amgen both have Treg-selective IL-2 agonists in Phase 2b trials, showing promise. Anokion is testing liver-targeted tolerance platforms in celiac disease. COUR licensed its nanoparticle approach to Takeda. Everyone, essentially, is hunting for the next mechanism, the next angle, the next way to calm overactive immune systems without simply carpet-bombing inflammation.
The need is real enough. A 2022 Mayo Clinic analysis of electronic health records found roughly 15.4 million Americans living with autoimmune conditions. Projections through 2040 show rising incidence across rheumatoid arthritis, inflammatory bowel disease, multiple sclerosis, psoriasis. The drugs are getting better, yes. But the problem is outrunning the solutions.
What Parasites Know That We Don't
Here's what ticks know: how to feed for a week without getting noticed.
Tick saliva is, from an immunological perspective, a marvel. It contains proteins called evasins—so named because they let the parasite evade detection. Evasins bind and neutralize chemokines, those signaling molecules that orchestrate immune cell recruitment. One protein family can target multiple inflammatory signals at once, a kind of molecular multitasking drug developers dream about but struggle to engineer.
There's precedent for turning this biology into medicine. Akari Therapeutics took a tick-derived complement inhibitor called OmCI and re-engineered it as nomacopan. It inhibits both C5 complement and the leukotriene LTB4—two separate inflammatory pathways. The drug has advanced through clinical trials in paroxysmal nocturnal hemoglobinuria, bullous pemphigoid, and several ocular indications. Not a blockbuster yet, but proof that parasite proteins can become human therapeutics.
Worms, meanwhile, have their own bag of tricks. Helminths—intestinal parasites that afflicted humans for most of our evolutionary history—secrete molecules that fundamentally rewire immune responses. They shift the balance from inflammatory Th1 and Th17 patterns toward regulatory Th2 and Treg states. Think of it as turning down the immune system's volume knob rather than hitting mute.
Heligmosomoides polygyrus, a mouse gut worm, produces proteins (called TGM proteins) that mimic transforming growth factor-beta (TGF-β). These mimics bind TGF-β receptors and co-receptors like CD44 and CD49d, inducing FOXP3+ regulatory T cells in human tissue samples. ES-62, a glycoprotein from filarial nematodes, shows anti-inflammatory effects in rodent models of rheumatoid arthritis and lupus. Researchers have even developed small-molecule analogues that disrupt MyD88 signaling—a critical node in innate immunity.
Viruses contribute yet another toolkit. Myxoma virus, which causes a lethal disease in European rabbits but is harmless to humans, produces Serp-1. It's a serine protease inhibitor (a serpin) with broad anti-inflammatory and anti-coagulant properties. Viron Therapeutics pushed Serp-1 into Phase 2a trials for acute coronary syndrome. Recent preclinical work has extended its potential to SARS-CoV-2 lung injury and colitis models.
These aren't obscure academic findings. They sit atop decades of literature. Microbiologist Graham Rook coined the "Old Friends" hypothesis to explain modern immune dysregulation: we've lost contact with co-evolved organisms that historically trained our immune systems. Parasites, in this framework, aren't just pathogens. They're regulators we've discarded, perhaps prematurely.
The question—the one Ditto Biosciences is betting on—is whether we can harvest the molecular mechanisms without reintroducing the organism itself. Live helminth therapy, tested in Crohn's disease patients using Trichuris suis ova (pig whipworm eggs), produced mixed results at best. A large randomized controlled trial failed its primary endpoint. Turns out the therapeutic window for swallowing worms is narrower than early enthusiasm suggested.
Better, perhaps, to identify the active ingredients and engineer them for safety and scale.
From PhD Labs to YC Demo Day
Ditto Biosciences was incorporated in California in September 2025. The founding team brings the kind of credentials investors like to see on slide decks, though how those credentials translate to commercial success remains, as always, an open question.
Adair Borges, the scientific lead, holds a PhD in parasitology, virology, and genomics from UCSF. She's published over 50 papers and co-invented patents on anti-CRISPR biology—work exploring how bacteriophages suppress bacterial immune systems. Dennis Sun, with stops at Harvard and UC Berkeley, focuses on product development and partnership strategy. Emily Weiss, PhD, spent time at Illumina and DuPont before joining Ditto to lead the platform for engineering parasite-derived therapeutics.
Academic pedigree doesn't guarantee success in biotech, but it opens doors. Y Combinator's backing signals something—not validation, exactly, but enough credibility to warrant attention. And the pitch, at least, is coherent.
Ditto's approach layers evolutionary biology with machine learning. Start with proteins parasites deploy during human infections—molecules that, by definition, already work in a biological context. Then apply structure prediction tools like AlphaFold 2, which solved the protein-folding problem and effectively gave researchers a cheat code for annotating unknown sequences. Train AI models on binding data to predict which candidates will hit specific human targets: cytokine receptors, chemokine pathways, complement proteins, innate immune sensors.
The company claims to have screened more than one million parasite proteins from viruses, ticks, and worms. Thousands, they say, show predicted activity against clinically validated targets. Early binding assays reportedly show affinities in the 1–2 nanomolar range—competitive with engineered monoclonal antibodies, if the data holds.
They're also building something less conventional: a human tissue biobank designed to map patterns of immune memory against parasite antigens. The goal is to predict immunogenicity—the body's tendency to recognize and attack foreign proteins—before a candidate ever reaches clinical testing. It's a practical response to what's historically been a deal-breaker for non-human therapeutics.
The Immunogenicity Problem (Or, Why Foreign Proteins Get Attacked)
Non-human proteins face a brutal reality in human bodies. The immune system evolved to recognize "non-self" and eliminate it. That's great when the target is a virus. Less great when it's your therapeutic.
Anti-drug antibodies (ADAs) can neutralize a drug's effect, reduce its half-life, or trigger adverse reactions ranging from mild to severe. Both FDA guidance (2014) and EMA guidance (2017) outline requirements for assessing immunogenicity in therapeutic protein development. The regulatory bar is high because the clinical risk is real.
The standard industry response is "de-immunization"—identifying and removing T-cell epitopes, the stretches of amino acids that bind MHC molecules and activate T cells. Companies like Abzena offer platforms (Composite Proteins, EpiScreen) that computationally predict epitopes and validate them in human T-cell assays. The tools have improved, but prediction remains imperfect. Clinical monitoring remains the gold standard.
Nomacopan offers a template, though not necessarily an encouraging one. The tick-derived protein was recombinantly expressed, humanized to some degree, and tested across multiple trials. Immunogenicity profiles were manageable, but the drug hasn't become a commercial juggernaut. Clinical success with parasite proteins is possible. Guaranteed, it is not.
Ditto's tissue biobank represents a bet that better data upfront—mapping human immune memory against parasite antigens—can improve the odds. Whether that's sufficient remains to be seen. Protein engineering is iterative, expensive, and littered with candidates that looked promising in vitro but failed in humans.
AI as Accelerant (Maybe)
AlphaFold's impact on structural biology is well-documented. DeepMind's tool, named Nature's Method of the Year in 2021, effectively solved a 50-year-old problem: predicting a protein's 3D structure from its amino acid sequence. The AlphaFold database now annotates much of the human proteome and extends to microbial and parasite genomes.
This has changed the economics of early-stage drug discovery. What would have required years of crystallography can now be approximated computationally in hours. A 2025 review in the Annual Review of Pharmacology estimated that AI could reduce preclinical timelines and costs by 35–50%, though the authors cautioned that success depends heavily on data quality and validation rigor.
Ditto's claim to have screened over a million proteins would have been impractical a decade ago. Now it's a tractable, if ambitious, computational exercise. Whether those predictions translate to clinical efficacy is another matter entirely.
AI-driven drug discovery startups have proliferated in recent years. The clinical track record so far is mixed. For every success story, there are half a dozen candidates that looked great in silico but failed to demonstrate meaningful activity in humans. Biology remains stubbornly resistant to purely computational approaches. The wet lab still matters.
A 2020 analysis in the Journal of Natural Products found that 26–32% of new chemical entities approved between 1981 and 2019 were derived from or inspired by natural products. Venom-derived drugs like exenatide (from Gila monster saliva), eptifibatide (from rattlesnake disintegrin), and ziconotide (from cone snail toxin) prove the model works. Parasite proteins represent adjacent territory—explored but not exhaustively mined.
What Happens Next
Ditto faces the standard biotech gauntlet. Hit validation. Lead optimization. IND-enabling studies. Phase 1 safety. Phase 2 proof-of-concept. Each stage filters ruthlessly. Most candidates don't make it.
The company is early. Very early. Founded less than a year ago, it has binding data, a candidate database, and a tissue biobank under construction. No preclinical development candidates have been publicly disclosed. No IND filings. No partnerships. Just a premise, some data, and YC's backing.
But the timing might be right. The autoimmune market is growing, consolidating around novel mechanisms, and rewarding differentiation. AbbVie's recovery with Skyrizi and Rinvoq—both mechanistically distinct from Humira—illustrates how the market values innovation. Eli Lilly's $2 billion Repertoire deal signals ongoing appetite for next-generation platforms.
Parasite-derived proteins could slot into this emerging toolkit. Treg-biased therapies—like the IL-2 pathway agonists from Nektar and Amgen, both in Phase 2b—aim to restore immune balance rather than simply suppress inflammation. CAR-Treg cell therapies are entering early trials. Tolerance-inducing nanoparticles (Anokion, COUR) target antigen-presenting cells to reprogram immune memory.
A TGF-β mimic that selectively induces regulatory T cells without systemic immunosuppression. A chemokine-binding protein that neutralizes multiple inflammatory signals simultaneously. A viral serpin that interrupts the crosstalk between coagulation and inflammation. Each represents a path biology has already validated through millennia of host-parasite coevolution.
That validation is worth something—maybe not enough, but something. Fifteen million Americans with autoimmune diseases. A $72 billion market growing to $92 billion. Pricing pressure from Medicare and biosimilars forcing differentiation. It's a landscape where novel mechanisms matter, where companies are taking bigger scientific risks, where the old playbook (me-too antibodies with marginal improvements) is losing its luster.
The One-Million-Protein Question
Ditto's scale is its calling card. One million proteins, winnowed to thousands of candidates, further filtered to—presumably—a handful of leads worth developing. It's a brute-force approach to evolutionary biology, made tractable by AI and modern computational infrastructure.
Whether that translates to drugs depends on execution, biology, and a fair amount of luck. Parasitology offers a rich but risky starting point. The precedents—nomacopan, Serp-1, ES-62 derivatives—show it's possible but haven't yielded blockbusters yet. Immunogenicity, manufacturability, competitive positioning, clinical efficacy: each represents a potential failure mode.
Manufacturing protein therapeutics at scale requires solving chemistry, manufacturing, and controls (CMC) challenges that biotech veterans know well. Expression systems. Purification. Formulation stability. Regulatory pathways for novel biologics are well-established, at least. The FDA has seen plenty of recombinant proteins. Whether parasite-derived molecules can achieve the safety, efficacy, and commercial profiles needed to compete with established modalities remains an open question.
Perhaps the telling detail is the filtering itself. From a million to thousands to a few. It's a funnel that acknowledges how brutally selective drug development is—and how much computational power you can throw at the early stages before the real work begins.
For now, Ditto is a bet. A bet that parasites have spent long enough perfecting immune manipulation that their molecular tools are worth systematically harvesting. A bet that AI can meaningfully accelerate the discovery process. A bet that the autoimmune market is hungry enough for novel mechanisms that investors, partners, and patients will give parasite-derived proteins a fair shot.
The company is six months old. The data is preliminary. The platform is unproven in humans. But in a market desperate for new approaches, and equipped with tools that would have seemed like science fiction a decade ago, maybe—just maybe—the time has come to take evolution's experiments seriously.
Or maybe not. Biology, as every drug developer eventually learns, has a way of humbling even the most elegant theories. Ask again in five years.
