PREMIX: PRivacy-preserving EstiMation of Individual admiXture

Area of Science:

• Genomics
• Computational Biology
• Privacy-Enhancing Technologies

Background:

• Genomic data analysis often involves sensitive information, necessitating robust privacy measures.
• Collaborative analysis across multiple sites can enhance the accuracy and scope of genomic studies.
• Existing methods may not adequately protect individual genetic data during collaborative research.

Purpose of the Study:

• To propose a novel framework, PRivacy-preserving EstiMation of Individual admiXture (PREMIX), for secure collaborative estimation of individual genetic admixture.
• To leverage Intel Software Guard Extensions (SGX) for secure computation over confidential genomic data.
• To ensure the privacy and security of sensitive genomic data during admixture estimation.

Main Methods:

• Implementation of a feature selection module to identify the most informative Single Nucleotide Polymorphisms (SNPs).
• Utilizing an Expectation Maximization (EM)-based Maximum Likelihood estimator for individual admixture identification.
• Conducting all sensitive genomic data operations within a secure enclave provided by Intel SGX.

Main Results:

• Experimental results from simulations and the 1000 Genomes dataset demonstrate the efficiency of the PREMIX framework.
• The proposed framework achieves high accuracy in estimating individual genetic admixture.
• PREMIX effectively ensures a high level of security for genomic data operations.

Conclusions:

• PREMIX provides a secure and efficient solution for collaborative individual admixture estimation.
• The framework's reliance on Intel SGX guarantees the confidentiality of sensitive genomic data.
• PREMIX facilitates privacy-preserving genomic data analysis in multi-site collaborations.