Privacy-preserving Chi-squared testing for genome SNP databases.

Area of Science:

• Genetics
• Bioinformatics
• Cryptography

Background:

• Genome-wide association studies (GWAS) identify genetic variants linked to diseases.
• Chi-squared testing is commonly used in GWAS, but publishing results can risk privacy.
• Existing privacy methods for chi-squared values are limited to small datasets.

Purpose of the Study:

• To develop novel anonymization techniques for chi-squared values in GWAS.
• To address privacy leakage concerns in genetic association studies.
• To propose methods applicable to larger datasets without significant information loss.

Main Methods:

• Introduction of Rand-Chi and RandChiDist, novel anonymization algorithms.
• Experimental evaluation of the proposed methods using real-world genetic datasets.
• Comparison with existing differential privacy techniques for chi-squared values.

Main Results:

• Rand-Chi and RandChiDist demonstrate effective anonymization of chi-squared values.
• The proposed methods maintain data utility for larger contingency tables.
• Experimental results validate the efficacy of the novel approaches.

Conclusions:

• Rand-Chi and RandChiDist offer improved privacy-preserving solutions for GWAS.
• These methods overcome limitations of previous anonymization techniques.
• The study contributes to secure genetic data analysis and disease research.