This contains Python codes used in our experiments on differentially private (DP) GWAS statistics based on their
This study focuses on key statistics in GWAS:
"Accuracy" folder contains the experimental results on differences between the original and DP statistics.
In addition to the results provided in our paper, those for smaller (
(Please also see Important Notes and Errata for the details of the experiments.)
"RunTime" folder contains the results on the execution time to compute the
The procedure to generate simulation data for the above experiments can be found in "SimulationData" folder.
"StatsFeature" folder contains the codes for analyses on the characteristics of each statistics. The detailed explanation and discussion are provided in Section IV of the main document.
In our experiments on
・In this study, the discussion for the
・The experimental results (regarding the analysis on
・In practice, after determining the range of the number of cases and controls for the analyses, it would be recommended to construct a specific algorithm for computing
・As for the cases when the number of cases and controls are far apart, the
・Although the analyses and discussions in Section IV are just simple examples of the procedures for constructing algorithms to calculate
(・Note that the evaluations for the TDT statistics are rigours.)
・Completely eliminating restrictions on the number of cases and controls for the
・Covering all statistical analysis in GWAS, e.g., Cochran-Armitage trend test and EIGENSTRAT.
・Improving the selection of random variables in Algorithm 1. / Would the idea of comparing multiple random variables also work for cases other than when using the standard Cauchy distribution?
・Investigating and analyzing other important characteristics and possibilities of
・Developing (efficient) methods for publishing
・Developing highly accurate methods for publishing top
・Constructing a more general and efficient algorithm to compute
・Considering
← In fact, the existing
For details of our methods, please see our paper entitled "Privacy-Preserving Publication of GWAS statistics using Smooth Sensitivity" (https://doi.org/10.1109/PST58708.2023.10320160) presented at PST 2023.
Errata:
・The last sentence in Section III.B.1. should be
・Section III.B.1. "The
・Section III.B.1.
・Section III.B.1.
・Figs. 5 and 6, The labels of
Akito Yamamoto
Division of Medical Data Informatics, Human Genome Center,
the Institute of Medical Science, the University of Tokyo