Privacy-preserving genotype imputation with fully homomorphic encryption.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Genotype imputation is crucial for understanding phenotype-genotype relationships, aiding in QTL mapping and GWAS.
• Large genomic datasets present computational and storage challenges for local servers.
• Cloud-based solutions for genotype imputation raise significant privacy concerns regarding sensitive genetic data.

Purpose of the Study:

• To develop an efficient and privacy-preserving algorithm for genotype imputation.
• To address the privacy risks associated with cloud-based genomic data analysis.
• To provide a scalable solution for real-world genotype imputation applications.

Main Methods:

• Developed p-Impute, a novel algorithm utilizing homomorphic encryption for privacy-preserving computations.
• Implemented a k-nearest neighbor-like approach for inferring missing genotypes within genomic blocks.
• Ensured computations are performed on encrypted data, preventing cloud-based decryption of private genotypes.

Main Results:

• p-Impute achieves accuracy comparable to state-of-the-art plaintext genotype imputation methods.
• The algorithm demonstrates linear scalability in memory and time requirements with an increasing number of samples.
• Successfully enables secure, cloud-based genotype imputation without compromising data privacy.

Conclusions:

• p-Impute provides an effective solution for privacy-preserving genotype imputation in cloud environments.
• The method enhances the feasibility of large-scale genomic analyses by mitigating privacy risks.
• p-Impute is a scalable and accurate tool for researchers working with sensitive genomic data.