A community assessment of privacy preserving techniques for human genomes

Table 1 Results of Task 1.

		Baseline		Team 1	Team 2	Team 3	# of sig SNVs
		SNV-based	Haplotype-based	U Oklahoma	UT Dallas	McGill U
D1	Power Cutoff 5.00E-02 1.00E-03 1.00E-05	0.05 TPR. FPR 0.864. 0.844 0.632. 0.774 0.643. 0.700	0.03 TPR. FPR 0.910. 0.612 1.000. 0.493 1.000. 0.475	0.61 TPR. FPR 1.000. 0.941 1.000. 0.884 1.000. 0.879	0.04 TPR. FPR 1.000. 0.855 1.000. 0.791 1.000. 0.737	0.01 TPR. FPR 1.000. 0.886 1.000. 0.798 1.000. 0.737	22 19 14
D2	Power Cutoff 5.00E-02 1.00E-03 1.00E-05	0.04 TPR. FPR 0.933. 0.924 0.800. 0.862 0.625. 0.788	0.115 TPR. FPR 0.978. 0.804 1.000. 0.708 1.000. 0.504	0.005 TPR. FPR 1.000. 0.958 1.000. 0.909 1.000. 0.876	0.01 TPR. FPR 0.533. 0 1.000. 0 1.000. 0	0.09 TPR. FPR 0.956. 0.746 1.000. 0.582 1.000. 0.425	45 15 8

In the first column, D1 refers to 200 participants, 311 SNVs (~29504091-30044866, chr2) and D2 refers to 200 participants, 610 SNVs (~55127312-56292137, chr10). The rows labeled 'Power' indicate the ratio of identifiable individuals using the likelihood ratio test in the case group. The other rows start with a cutoff threshold for the χ²test (e.g., 5 × 10^-², 10^-³, 10^-⁵), for which two measurements (true positive rate and false positive rate for SNVs using the χ² test) were calculated under each method. The last column corresponds to the number of significant SNVs calculated using the original data (i.e., without added noise).

ISSN: 1472-6947