Towards a new standard in genomic data privacy: a realization of owner-governance

Jingcheng Zhang, Yingxuan Ren, Man Ho Au, Ka-Ho Chow, Yekai Zhou, Jason Li Chen, Yanmin Zhao, Junhao Su, Ruibang LuobioRxiv 2024 · DOI

TL;DR

This paper introduces Governome, a novel owner-governed system for genomic data management that prioritizes individual control, data privacy, and verifiability. By integrating blockchain, homomorphic encryption, and zero-knowledge proofs, Governome empowers individuals with absolute control over their genomic data throughout its lifecycle, enabling secure and transparent data sharing for research and analysis.

What Problem Does This Paper Address?

The rapid advancements in genomic sequencing technologies have led to an explosion of personal genomic data. However, current data management systems fall short in protecting individual privacy and granting users sufficient control over their sensitive genetic information. This inadequacy hinders data sharing for crucial biomedical research and leaves individuals vulnerable to potential misuse or mishandling of their data by third-party services. Existing solutions often struggle with issues like data access revocation, re-identification risks, and a lack of transparency in data usage. The paper aims to bridge this gap by proposing a system that ensures owner authority, lifecycle data encryption, and verifiability, creating a new standard for genomic data privacy.

Methods

Governome is built upon three core properties of owner-governance: Owner Authority (OA), Lifecycle Data Encryption (LDE), and Verifiability (VER). To achieve these, the system employs a multi-layered architecture:

  • Consensus Layer (Blockchain): This layer utilizes a blockchain to manage user agreements, permissions, and transactions. It ensures decentralized control, dynamic permission management, and transparent auditability of data usage. Zero-knowledge proofs are used here to enforce data integrity.
  • Computing Layer (Cryptographic Techniques): This layer handles secure computation on encrypted data. It leverages:
  • Homomorphic Encryption (HE): Specifically, a third-generation HE technique (Torus) is used to perform computations directly on encrypted genomic data, ensuring that no plaintext data is ever exposed during analysis.
  • Stream Cipher: Used for initial encryption of genomic data to maintain efficiency in storage and on-the-fly conversion to HE ciphertext for computation.
  • Secure Multi-party Computation (SMPC) and Threshold Fully Homomorphic Encryption (ThFHE): These are employed for collaborative generation of HE keys and decryption of results, ensuring that no single party holds the complete key, thus preventing unauthorized access.
  • Application Layer: This layer provides a user-friendly interface for interacting with the system, defining genomic analysis tasks, and managing user roles.

Datasets

The primary dataset used for evaluation in this paper is the 1000 Genomes Project (1kGP) dataset, which comprises whole-genome variants of 2,504 individuals. Mock-up phenotypes for these individuals were provided by the Hail library.

Key Insights and Ideas

  • Owner-Governance as the Next Frontier: The paper argues that true genomic data privacy and control can only be achieved through an owner-governed system, moving beyond current centralized or permissioned models.
  • Synergy of Blockchain and Cryptography: The successful implementation of Governome hinges on the synergistic combination of blockchain for decentralized control and auditability, and advanced cryptographic techniques (HE, ZKP) for robust privacy and verifiability.
  • Lifecycle Data Encryption is Non-Negotiable: The concept of LDE is central, emphasizing that genomic data must remain encrypted from storage through computation, even from the data owner, to prevent any risk of leakage or misuse.
  • Verifiability for Trust: Ensuring that both data integrity and computation results are verifiable is crucial for building trust in a decentralized system where multiple parties are involved.
  • Practicality of HE for Complex Tasks: The use of third-generation HE techniques demonstrates the feasibility of performing complex genomic analyses (like GWAS) on encrypted data, overcoming limitations of earlier HE schemes.

Key Results

  • Efficient Querying: A query for a variant's genotype distribution among 2,504 1kGP individuals can be completed in under 18 hours on an ordinary server, with data conversion taking approximately 13 hours and data analysis taking about 4.5 hours.
  • Scalability: The performance of cohort studies scales linearly with the number of samples, indicating Governome's potential to manage large population-scale genomic projects.
  • Low Owner Overhead: For data owners, generating an access token proof requires minimal memory consumption (around 1GB) and takes about a minute, making it feasible for typical personal devices.
  • Comprehensive Owner Control: Governome successfully fulfills the three properties of owner-governance: Owner Authority, Lifecycle Data Encryption, and Verifiability, offering dynamic permission management and full transparency to data owners.
  • Support for Diverse Analysis Tasks: The system supports individual variant queries, cohort studies, GWAS analysis, and forensics, catering to different user groups (data owners, authorities, research entities).

Limitations and Future Work

  • Variant Indexing: The current implementation primarily supports rsID as the variant index, which is insufficient for representing all variants in a whole-genome sequenced genome. Future work will involve supporting VCF format for easier storage of whole-genome data.
  • Data Storage: Governome currently stores only genomic data and relies on external institutions for demographic and phenotypic data. While this design isolates critical data types, future work could explore integrating the storage of demographic and phenotypic data.
  • Quantum Resistance: Governome is not currently quantum-resistant. The authors plan to explore optimized cryptographic methods and privacy protocols to achieve post-quantum reliability.
  • Efficiency of HE: While third-generation HE is used, computations involving complex operations or many branches can still be computationally intensive. Further optimization of HE schemes and algorithms is an ongoing area of research.

Summary updated

Abstract

Abstract With the rapid developments in sequencing technologies, individuals now have unprecedented access to their genomic data. However, existing data management systems or protocols are inadequate for protecting privacy, limiting individuals’ control over their genomic information, hindering data sharing, and posing a challenge for biomedical research. To fill the gap, an owner-governed system that fulfills owner authority, lifecycle data encryption, and verifiability at the same time is prompted. In this paper, we realized Governome, an owner-governed data management system designed to empower individuals with absolute control over their genomic data during data sharing. Governome uses a blockchain to manage all transactions and permissions, enabling data owners with dynamic permission management and to be fully informed about every data usage. It uses homomorphic encryption and zero-knowledge proofs to enable genomic data storage and computation in an encrypted and verifiable form for its whole lifecycle. Governome supports genomic analysis tasks, including individual variant query, cohort study, GWAS analysis, and forensics. Query of a variant’s genotype distribution among 2,504 1kGP individuals in Governome can be efficiently completed in under 18 hours on an ordinary server. Governome is an open-source project available at https://github.com/HKU-BAL/Governome .
BibTeX
@misc{zhang2024towards,
  author = {Jingcheng Zhang and Yingxuan Ren and Man Ho Au and Ka-Ho Chow and Yekai Zhou and Jason Li Chen and Yanmin Zhao and Junhao Su and Ruibang Luo},
  title = {{Towards a new standard in genomic data privacy: a realization of owner-governance}},
  year = {2024},
  doi = {10.1101/2024.07.23.604393},
  howpublished = {bioRxiv},
}
Figure