SecureGenomics™

Compute on encrypted data. Mathematical guaranties on individual data privacy.

SecureGenomics Engine is a platform for privacy-preserving genomic analysis using Fully Homomorphic Encryption (FHE) and federated computing. It lets scientists run population-scale studies, GWAS, allele frequency analysis — all without ever decrypting sensitive data.

   

Built for

• 🧪 Biobanks — monetize datasets without compromising privacy
• 🧠 Researchers — collaborate across silos, globally, securely
• 🌍 GDPR/HIPAA - safe by design
• 🔐 Zero-trust compute with cryptographic guarantees

   

⚠️ Alpha stage — active research tool. Contributions & collaborations welcome.

   

Quick Start

Install

git clone https://github.com/securegenomics/securegenomics.git && cd securegenomics && bash setup.sh

🚀 Super Simple Workflow

For those who want the quickest path to running secure genomic analysis

For researchers:

# 1. Login
$ secgen login

# 2. Create project (interactive - choose protocol)
$ secgen create

# 3. Generate crypto keys
$ secgen keygen <project-id>

# 5. Run analysis
$ secgen run <project-id>

# 6. Check status
$ secgen status <project-id>

# 7. Get results
$ secgen result <project-id>

For data owners (biobanks, individuals, etc.):

# 1. Upload data
$ secgen upload <project-id> <data.vcf>

download example human genome to try out quickly

$ mkdir -p ~/data/genome && wget -P ~/data/genome https://storage.googleapis.com/genomics-public-data/simons-genome-diversity-project/vcf/LP6005441-DNA_C01.annotated.nh2.variants.vcf.gz && gunzip ~/data/genome/LP6005441-DNA_C01.annotated.nh2.variants.vcf.gz

🚀 How it works? - explained on a common scenario

Bob (scientist) 👨

On his laptop 💻

# Bob creates a new project
$ secgen create
# ☝️ this command, asks Bob to choose an open-source, shareable experiment protocol from https://github.com/securegenomics/ . He chooses `protocol-alzheimers-sensitive-allele-frequency`. All protocols involve scripts for encoding, encryption, computation, decoding, and result interpretation

# Bob generates a public-private crypto context pair
$ secgen keygen <project-id> 
# ☝️ this command, under the hood, uploads public crypto context to the SecureGenomics server

Alice (owns sensitive data) 👩

On her computer 🖥️

# 👨 – Hey Alice, can you contribute to my new experiment with your DNA?
# 👩 – Sure, I love science! But, but I also love my privacy :(
# 👨 – Don't worry, I know an awesome secure tool to do this! Use my <project-id>, encrypt your data and upload to the server!
# 👩 – Cool!

# Alice uploads her genomic data using the complete pipeline
$ secgen upload <project-id> data.vcf
# ☝️ under the hood, it encodes, encrypts, and uploads the data in one command

# Or Alice can do it step by step:
# $ secgen encode <project-id> data.vcf
# $ secgen encrypt <project-id> data.vcf.encoded 
# $ secgen data upload <project-id> data.vcf.encrypted

# ℹ️ All above commands use the online protocol code from shared experiment Github repository.

Others (own sensitive data too)

On their local computers 💻

# Dave, Frank, George, Carol, ... all do the same:
$ secgen upload <project-id> their-data.vcf
# ☝️ Each person uploads their encrypted genomic data to the same project

Same Bob again (the scientist) 👨

On his laptop 💻

# Checks his project, and sees all his friends uploaded– 100s of encrypted genomes! 
$ secgen view <project-id>

# Bob now runs the experiment
$ secgen run <project-id>
# ☝️ FHE computation, as described in the protocol, is performed on the server.

# After, he downloads and decrypt experiment results with his private key
$ secgen result <project-id>

What really happened?

• 🙋‍♂️ Bob is happy, because he did an analysis on lots of people's DNA
• 🙋‍♀️ Alice and other contributors are happy, because they kept their DNA private (cryptographically guaranteed)
• 🗄️🔐 Data was always in encrypted form on the server

Experiment Protocols

Main Hub - github.com/orgs/securegenomics/repositories

Pick a research protocol above, or create your custom protocol and merge into this repo.

This is the truth base for all computations. You can verify and prove others which computation script was used in your experiment.

Hyper-sharable, cryptographically verifiable science.

Resources

• docs/guide.md
  • for users – installation & commands
• docs/design.md
  • for developers
• github.com/barisozmen/genomic-privacy-book/
  • Categorization of genomic privacy concerns (see)
  • Private vs Public Genomic Data (see)
  • FHE mathematical foundations (fhe, math overview, algebra, lattice-based cryptography)
  • Privacy technologies overview (see)

Future Work

🔐 Hash-Based Provenance: Scientific Truth as a Chain of Commitments

Treat every computation as a cryptographically signed step:

• Every input dataset has a hash
• Every version of your code has a hash
• Each computation step generates a new hash by combining:
  • Data hash
  • Code hash
  • Config/parameter hash
  • Previous step hash (for lineage)

This results in a chain of composable, tamper-proof proofs that describe exactly how an output came to be — like Git commits, but across data + code + math.

📌 Even if you don’t share your input data, others can verify that your output hash is consistent with what it should be, given the hash of your data and your open-source computation protocol.