Description

The increasing availability of genomic data has transformed biomedical research, enabling large-scale studies of genetic diseases, personalized medicine, and population health. At the same time, genomic information is among the most sensitive forms of personal data. Individuals may be unwilling or legally unable to share their genomes with external organizations, creating a fundamental tension between data privacy and scientific collaboration.

The module Multi-Party Computations in Genomics explores how modern cryptographic techniques make it possible to perform joint computations on sensitive genomic datasets without revealing the underlying private data. Students will learn the principles of Secure Multi-Party Computation (MPC) and its role within the broader landscape of privacy-preserving machine learning, comparing it with approaches such as homomorphic encryption and zero-knowledge proofs. The course demonstrates how these techniques enable collaborative genome-wide association studies (GWAS), polygenic risk score (PRS) computation, and secure statistical analysis across multiple institutions while maintaining strong privacy guarantees.

Beyond cryptographic foundations, the module introduces the structure and characteristics of genomic data, including Variant Call Format (VCF), single nucleotide polymorphisms (SNPs), and the statistical methods commonly used in computational genomics. Students will gain an understanding of the computational challenges posed by modern genomic datasets, including their scale, high dimensionality, and distributed nature. The course also discusses practical MPC frameworks and architectures, performance trade-offs, and real-world deployment considerations for secure collaborative analytics.

The module consists of lectures and practical sessions. The lectures introduce the theoretical foundations of genomic data analysis and secure computation, while the practical sessions familiarize students with contemporary MPC frameworks and demonstrate how privacy-preserving genomic analyses can be implemented in practice. Throughout the course, students work with realistic genomic datasets and learn how secure computation protocols are applied to solve practical biomedical problems.

In summary, the module pursues the following learning objectives:

• Students understand the fundamentals of genomic data, including SNPs, VCF files, GWAS, and polygenic risk scores.
• Students are familiar with the principles of Secure Multi-Party Computation and its security guarantees.
• Students can compare MPC with other privacy-preserving technologies, including homomorphic encryption and zero-knowledge proofs, and understand their respective strengths and limitations.
• Students understand how secure computation enables collaborative genomic analysis without exposing sensitive patient data.
• Students gain practical experience with modern MPC frameworks and learn how to implement privacy-preserving statistical and machine learning workflows on distributed genomic datasets.
• Students appreciate the computational, ethical, and regulatory challenges associated with secure genomic analytics and personalized medicine.

The course assumes basic knowledge of probability, statistics, and linear algebra. Prior exposure to programming in Python is beneficial but not strictly required, as practical examples and implementation details will be introduced throughout the module. Students with an interest in machine learning, cryptography, biomedical data science, or privacy-enhancing technologies will find the course particularly valuable.

We look forward to welcoming you to Multi-Party Computations in Genomics.