ขอเชิญเข้าฟังบรรยายในหัวข้อ Privacy ‘in Use’: Securing Genome Analysis in Untrusted Clouds with Trusted Execution Environments
22 November 2024
ขอเชิญเข้าฟังบรรยายในหัวข้อ
Privacy ‘in Use’: Securing Genome Analysis in Untrusted Clouds with Trusted Execution Environments
โดย
Natnatee Dokmai, PhD
Postdoctoral Associate Department of Biomedical Informatics and Data Science, Yale School of Medicine https://medicine.yale.edu/profile/natnatee-dokmai/
เวลา 10:00 - 11:00
ผ่าน ZOOM
Zoom link: https://cmu-th.zoom.us/j/93582832416
ABSTRACT
Conventional cloud security measures safeguard data ‘at rest’ and ‘in transit’ but fail to protect data ‘in use’, leaving sensitive data vulnerable during runtime analysis. This vulnerability is especially critical in genotype imputation services, where researchers upload low-resolution genomic data to public servers to statistically infer missing genotypes using high-quality reference datasets. While genotype imputation is essential for downstream analyses, such as gene-disease association studies, it introduces significant privacy risks for sensitive and legally protected genomic data, potentially deterring researchers and restricting access to these valuable services.
In this talk, I will introduce Trusted Execution Environment (TEE) technologies—a hardware-based solution available in commercial server-grade processors—to secure data ‘in use’ within untrusted cloud environments. I will then present our design of a secure genotype imputation service leveraging TEE technologies. Our approach addresses TEE architectural limitations by incorporating algorithmic hardening against side-channel attacks and optimizing performance to mitigate computational overhead. The resulting tool delivers accuracy and efficiency comparable to state-of-the-art imputation tools while ensuring robust in-use data security. This work highlights the feasibility of secure, privacy-preserving genomic analysis in untrusted cloud settings and lays the groundwork for broader applications in biomedical research.