Skip to main content

Table 3 A summary of the results for the secure outsourcing tasks

From: Protecting genomic data analytics in the cloud: state of the art and opportunities

Task 1.1   MAF \( {\chi}^2 \)
311 SNPs 610 SNPs 311 SNPs 610 SNPs
Time Memory Time Memory Time Memory Time Memory
Microsoft 17.44 s 130.48 MB 26.31 s 247.30 MB 16.88 s 118.08 MB 27.11 s 234.73 MB
UCIa 0.589 s 3.320 MB 0.886 s 3.320 MB 0.659 s 3.320 MB 0.871 s 3.320 MB
Stanford/MIT 1.069 s 8.0 MB 1.847 s 13.0 MB 1.069 s 8.0 MB 1.847 s 13.0 MB
U of Tsukuba 55.20s 31.81 MB 112.32 s 32.67 MB 55.21 s 31.81 MB 112.32 s 32.67 MB
Task 1.2 (a)   5 k 10 k 100 k
Accuracy Time Memory Accuracy Time Memory Accuracy Time Memory
Plaintext data 3099 0.076 s 1.64 MB 3306 0.118 s 2.43 MB 134252 134252 13.52 MB
IBM 3099 79.4 s 1.416 GB 3306 86.8 s 1.419 GB 134260 134260 2.168 G
Microsoft 3099 44.664 s 513.7 MB 3306 80.031 s 720.5 MB    
Stanford/MIT 3082 20m37s 2.77 GB 3275 36m27s 4.03 GB 132703 132703 7.50 GB
Task 1.2 (b)   5 k 10 k 100 k
Accuracy Time Memory Accuracy Time Memory Accuracy Time Memory
Plaintext data 9089 0.106 s 2.45 MB 16667 0.144 s 2.53 MB 191986 1.528 s 25.8 MB
IBMb 5328 91.7 s 1.42 GB 8318 106.3 s 1.45 GB 153266 555.2 s 2.29 GB
Microsoft 9089 91.09 s 701 MB 16665 181.92 s 1.29 GB    
  1. aThe algorithm encrypts local counts instead of input data for secure data outsourcing, and was disqualified in the competition
  2. bAn approximate algorithm (with about 22 % error), which was not considered in the competition
  3. Winners are in boldface