Secure outsourced computation of genomic data

项目摘要 在精准医疗时代，基因组数据正在与其他医疗数据整合 以支持个性化和校准的临床决策。基因组序列数据也是 电子健康记录（EHR）系统中存储的数据量很大，需要单独存储。 虽然云计算提供了一个具有成本效益和可扩展的平台， 对外包基因组数据的关切是具有挑战性的问题。普遍的看法是， 远程数据访问的便利性和隐私保护是相互矛盾的。 我们提出了一个新的基因组学存档和通信系统（GACS）， 通过使用最先进的同态加密算法和矩阵 数据和查询的表示。 在该系统中，变体被表示为向量，其通过 客户端并存储在GACS服务器上。当需要进行分析时，将在 矩阵的形式。该矩阵是加密的（或者可以根据任务保持明文）， 发送到GACS服务器。服务器对加密数据进行计算，产生加密结果 并将其返回给客户端，客户端具有解码它的密钥。GACS无法解密 数据或加密查询，从而保证隐私和安全得到维护 关于GACS算法的初步结果表明，解密后，结果是 这与在明文上计算的结果相同。在这个项目中，我们将实现我们的GACS 系统软件模块，并结合实例演示了系统的使用方法，从三个方面使用- 案例：药物基因组学、临床试验合格性和疾病风险分析。我们将 测量所有三种使用情形中的性能速度和内存消耗。 作为云托管服务的GACS系统可以减少医疗保健的计算负担 设施它可以为小型医疗机构提供相同的基因组分析能力 提供给大医院。此外，临床决策支持（CDS）可以部署在 GACS。随着临床指南的发展，以应对新发现的遗传变异， 疾病和药物，医疗机构可以保持遵守准则。

Project Summary In the age of precision medicine, genomic data are being integrated with other health care data to support personalized and calibrated clinical decision-making. Genomic sequence data are too large to be stored in electronic health record (EHR) systems and need to be separately stored. While cloud computing offers a cost-efficient and scalable platform, the privacy and security concerns about outsourcing genomic data are challenging issues. The common perception is that the ease of access to remote data and the protection of privacy are at odds with each other. We propose a new genomics archiving and communications system (GACS) that meets both requirements by using state-of-the-art homomorphic encryption algorithms and matrix representation of data and queries. In this system, variants are represented as vectors, that are homomorphically encrypted by a client and stored on the GACS server. When analysis is required, a query is generated in the form of a matrix. This matrix is encrypted (or can remain in plaintext depending on the task) and sent to the GACS server. The server computes on encrypted data, produces an encrypted result and returns it to the client, who has the secret key to decode it. The GACS is not able to decrypt the data or the encrypted queries, thus guaranteeing that privacy and security are maintained on the GACS. Preliminary results of the algorithms show that after decryption, the results are the same as results from computing on plaintext. In this project, we will implement our GACS system software modules and demonstrate the use of the system with examples from three use- cases: pharmacogenomics, clinical trials eligibility and analysis for disease risks. We will measure performance speed and memory consumption in all three use-cases. A GACS system as a cloud-hosted service can reduce the computational burden on healthcare facilities. It can provide small healthcare facilities with the same genomic analysis capability available to larger hospitals. In addition, clinical decision support (CDS) can be deployed on the GACS. As clinical guidelines evolve in response to new discoveries linking genetic variants to disease and medicines, healthcare facilities can stay in compliance with the guidelines.