ReadMe:

The program PrivMAF.py allows the user to make use of the privacy results presented the paper "Providing Provable Privacy Promises for Aggregate Genomic Data".

To run the program you need python installed, as well as numpy. No other special packages should be required.

In order to run the program type:

python PrivMAF.py [arguments]

where the arguments are described below.

You need to pass the program 3 arguments. The first argument is a .tped file that contains the genotype information from the participants in your study. Note that we assume that there are no unknown alleles.

The second argument is a file containing minor allele frequency information from the population from which your study is drawn. For each SNP you are interested in this file contains one line which lists the SNP's name followed by the minor allele frequency of that SNP in the background population, followed by the major allele (the major allele is one of A, G, T, or C). Note we assume all SNP have MAF above 5% (in order to ensure our approximations work).

The third argument is any integer larger than the size of your study population. This number if the size of the background population from which your study population was drawn (that is to say it is equal to the number of people who an adversary might reasonably suspect could have participated in the study).

Finally there is an optional fourth argument, a. This parameter is the cutoff threshold if you are using for ALGT. See the paper for more details. Note that this parameter needs to be a number between 0 and 1.

The program outputs to the screen. First it outputs the PrivMAF score calculated from your dataset. If you only passed the program 3 arguments the program then exists. If you passed it four arguments it then outputs two more results. First it outputs the rescaled threshold, beta, for ALGT (see paper for details). Finally it tells the user if the data is safe to release given a.


Example Of Use:

To get the privacy measure you could enter:

python PrivMAF.py study.tped mad.txt 1000 

python PrivMAF.py  study.tped  maf.txt  10000  .1

where the first command corresponds to just outputting the PrivMAF score, the second to running the entire ALGT algorithm. In the above maf.txt is the maf file, SNP.txt the list of SNP, study.tped the genotype data, 10000 the size of the background population, and .1 is the threshold a you want to use.