Genotype imputation estimates missing genotypes from the haplotype or genotype reference panel in individual genetic sequences, which boosts the potential of genome-wide association and is essential in genetic data analysis. However, the genetic sequences involve people's privacy, confirming an individual's identification and even disease information. This work proposes a secure genotype imputation model, which uses a linear regression model and the homomorphic encryption scheme over ciphertext to impute missing genotypes. The inference model is trained with float plaintext parameters, which are round into integers to avoid high complexity homomorphic evaluation on float number operations without bootstrapping operations. Even though the rounding parameters in the inference model are not the same as those in the trained model, We find that it will no effect on the outcome of the homomorphic prediction. Thus, a high-efficiency genotype imputation inference model over the ciphertext is obtained while keeping the high-security level. The simulation results indicate that the accuracy of the secure inference model is almost the same as the original model trained on float parameters. The secure inference model's accuracy is 98.6% for a single genotype.

Homomorphic encryption provides a way to perform deep learning over encrypted data and permits the user to encrypt the data before uploading, leaving the control of data on the user side. However, operations on encrypted data based on homomorphic encryption are time-consuming, especially in a deep convolutional neural network (CNN), which incorporates a large number of layers and operations. To speed up deep learning on encrypted data, we binarized the input data and weights of CNN model, while operations including the addition and multiplication in CNN become bit-wise operations. Therefore, the homomorphic evaluation of CNN can be performed in the binary field in a highly efficient way. We also construct an efficient pooling layer by designing circuits to perform comparison operations on the ciphertext. Simulation results clearly show that the convolution operation of the proposed model is at least 6.3 times faster than that of existing schemes. Last, our model exhibits no privacy leakage associated with the data being processed.