GitHub - popvax/yash: A personalized mRNA cancer vaccine for Yash, made at the RNA Foundry

GitHub

3 min read Original article ↗

rnafoundry.com/yash, manufactured at the RNA Foundry.

In January 2025, 32-year-old Yashdeep Bindal was diagnosed with stage IV pancreatic cancer. He had just returned to work after the birth of his daughter, Maya.

At the RNA Foundry, we are developing a personalized cancer vaccine for Yash. We’re sharing his tumour DNA, tumour RNA, matched-normal DNA and vaccine designs so that other researchers can help refine the work.

Lead candidate, shown over two rows: 5′ UTR, SEC, 24 antigen regions and linkers, MITD, stop codons, 3′ UTR and unspecified PolyA; the encoded protein is 724 amino acids.

Get the data · See the lead design · Run the code · Contribute

Data

Dataset When collected Collected by Files Size
Tumour DNA 15 Jan 2025 Caris BAM · index · Caris VCF 13.26 GB BAM
Tumour RNA 15 Jan 2025 Caris FASTQ R1 · FASTQ R2 · BAM · index 1.38 GB FASTQ pair; 1.32 GB BAM
Matched-normal DNA 30 Oct 2026 BAM · index 143.29 GB BAM
Gene expression 15 Jan 2025 Caris Supplied expression values 461.10 kB
HLA typing 15 Jan 2025 Caris Six class-I alleles used in our analyses 727 B

The data guide describes each dataset, sample and reference scope. The file catalog lists locations, sizes and SHA-256 checksums.

Clinical reports

Clinical timeline — diagnosis, treatments, scan findings and specimen collection.

See the lead design

Design #2 is the lead design, generated using matched-normal white-blood-cell DNA alongside tumour DNA and RNA. It contains 24 antigen regions in a 724-aa protein: 12 mutation, 5 splice, 5 CTA/self and 2 fusion regions.

PolyA and any downstream extension are unspecified; nucleotide files contain the known segments. Design #1, generated without matched-normal DNA, is retained for comparison.

Verify and reproduce

With Python 3.11 or 3.12, run from the repository root:

python3.11 -m venv .venv
source .venv/bin/activate
python -m pip install -e '.[mrna]'
yash-vaccine verify
yash-vaccine reproduce --output work/design-2-replay

These commands check and replay the lead design using bundled evidence and recorded predictions. They need no sequencing downloads or cloud account. The usage guide explains replay and alternative designs; the pipeline guide covers new raw-read processing and inference.

Make another candidate

Fork the repository, add your work under designs/your-design/, and submit a pull request. Include your rationale, data sources, code, commands and useful results. Annotated protein or RNA sequences are welcome; label unspecified components explicitly. An analysis or antigen list can be contributed before a complete construct is assembled.

See CONTRIBUTING.md. Questions and ideas are welcome as GitHub issues or at yash@rnafoundry.com.

Other directions to explore

This is one approach to vaccine design. Alternatives could examine MHC class II targets, immunogenicity, antigen trafficking, or linker optimization for processing and presentation.

Shared data are in data/, designs in designs/, maintained helpers in src/, and upstream jobs in workflow/. See the agent guide, reference acquisition guide, provenance records, required credits and retained-source notices for details.