Developing a cloud-based secure computation platform for genomics research

Abstract

Recent advances in genomic sequencing technologies and big data analytics present a golden opportunity for bioinformatics, making it possible to efficiently analyze hundreds of thousands of individual genomes and identify statistically significant genetic determinants of disease. However, most research institutes lack sufficient genomic data to detect fine-grained signals crucial for understanding complex human diseases, and data sharing is often impractical due to strict privacy protections. By leveraging a cryptographic technique known as secure multiparty computation (MPC), researchers can securely cooperate on large-scale genomic studies without revealing sensitive data to any collaborators. In this thesis, we propose a public cloud-based computational framework that implements MPC for an essential genomic analysis workflow known as genome-wide association study (GWAS). By productionizing secure GWAS tools in an easy-to-use interface that abstracts away the technical challenges involved with implementing and running a protocol on several independent, geographically separated machines, we hope to enable researchers around the world to launch meaningful genomic studies with minimal overhead. We hope such efforts will prove instrumental towards the broader aim of establishing a single general-purpose platform for secure genomics research.