VaccineOS

Peptide Results Page
Our Dashboard
Peptides associated to the mutatation

Inspiration

Every cancer is unique to the patient carrying it. We wanted to build a solution that assists clinicians in producing mRNA cancer vaccines, making their workflow more efficient, leading to more cancer patients treated. We were inspired by the promise of personalized mRNA cancer vaccines, and by Paul Conyngham, the Australian tech entrepreneur who made headlines in March 2026 after using ChatGPT and AlphaFold to design a custom mRNA cancer vaccine for his dying dog, Rosie. With no formal biology training, he sequenced Rosie's tumor DNA, used AI to identify mutations and predict neoantigen targets, and collaborated with UNSW's RNA Institute to manufacture the vaccine. Within weeks, her tumor shrunk by 75%. That story is proof that this pipeline can be accessible to anyone, not just research labs.

What it does

Vaccine OS is a full-stack neoantigen vaccine design platform. A clinician inputs a patient ID, cancer type, and HLA alleles, then uploads a VCF file containing the patient's tumor mutation data. The app parses the genomic variants, runs an automated peptide-suggestion pipeline, and returns a ranked list of neoantigen candidates personalized to that patient's immune profile, all viewable in a live processing dashboard. Vaccine OS still allows clinicians to utilize their medical instincts and knowledge, where they are able to make decisions on what peptides to consider, what neoantigen candidates to report, and the weights that affect our calculated scoring. It leads a clinician with the required information to send to the lab for mRNA vaccine development and testing.

How we built it

We built Vaccine OS on Next.js (App Router) with TypeScript and Tailwind CSS for the frontend. Supabase handles authentication, the database, and VCF file storage via signed upload URLs. The backend pipeline runs through Next.js API routes, orchestrating VCF parsing, peptide extraction, MHC binding prediction, and build logging in a sequential async flow.

Challenges we ran into

Discovering peptides results for mRNA vaccines require much more considerations that we originally thought, ranging from processing genomic files already integrated in clinical workflow and determining the considerations that characterizes tumor suppressing peptides. We spent over 12 hours doing the research necessary to comprehend each step necessary to take in a VCF file, parse it for mutations, find peptides, rank peptides, sequence peptides for mRNA vaccines, then present a vaccine that is structurally sound. This was very difficult.

Accomplishments that we're proud of

We shipped a working end-to-end pipeline, from raw genomic upload to ranked neoantigen output. The UI makes a deeply technical biomedical workflow feel approachable and clinical-grade. We turned a research process that takes several biologist, clinicians, and researchers several weeks to complete into an easy-to-use, integrable, and customizable software that only takes minutes, and allows mRNA vaccine testing documentation to be immediately sent to the lab.

What we learned

Building Vaccine OS required over 12 hours of deep research just to understand the pipeline end-to-end. We had to learn how to ingest and parse a VCF file for somatic mutations, extract candidate peptides from those mutations, apply ranking logic to identify the strongest neoantigen targets, sequence those peptides into a structurally viable mRNA construct, and finally present a vaccine design that is biologically sound. Every step unlocked a new layer of complexity we hadn't anticipated going in.

What's next for VaccineOS

What Integrate pMHC binding affinity scoring (via NetMHCpan or similar), add mRNA sequence optimization for the top-ranked peptides, and explore partnerships with oncology research labs for real patient data validation, taking what Conyngham proved possible and turning it into a repeatable clinical tool.