Abstract:
With the rise of consumer-centric genomic testing and the subsequent increase
in privacy concerns regarding genomic data theft, Fully Homomorphic Encryption
offers a viable approach to preserving data confidentiality while allowing computations
to be performed directly on encrypted data. This study explores the possible
implementation of a paternity test under FHE by developing a secure, privacypreserving
system that performs the genomic computations using the CKKS encryption
scheme and the TenSEAL library. Results show that using a large scaling
factor such as 260 yields very high accuracy (95%-100%) however there is a slight
tradeoff in the overall runtime. The system demonstrates that it is possible to perform
secure genomic computations while maintaining the accuracy and privacy of
sensitive genomic data.
