Compute Multi-trait GWAS with JASS

Once the GWAS summary statistics are integrated in the inittable, you can generate analysis for any combination and several joint tests with the command jass create-project-data (see command line usage for the detail of arguments).

Whatever the test used, the command will generate three output:

  • A HDFStore containing several tables (Each table can be read from the HDFStore with the pandas.read_hdf function): - 'SumStatTab' : The results of the joint analysis by SNPs - 'PhenoList' : the meta data of analysed GWAS - 'COV' : The H0 covariance used to perform joint analysis - 'GENCOV' (If present in the initTable): The genetic covariance as computed by the LDscore. - 'Regions' : Results of the joint analysis summarised by LD regions (Notably Lead SNPs by regions) - 'summaryTable': a double entry table summarizing the number of significant regions by test (univariate vs joint test)

  • A .png Manhattan plot of the joint test p-values:

_images/manhattan_glycemic_blood_asthma.png

  • A .png Quadrant plot which is a scatter plot of the minimum p-value by region of the joint test with respect to the minimum p-value by region of the univariate tests. This plot provides an easy way to see if your joint analysis detected association not previously reported in the litterature.

_images/quadrant_glycemic_blood_asthma.png

The Omnibus tests

If no method option, a Omnibus test analysis will be performed.

The SumZ test

If the flag --sumz is passed to the jass create-project-data a SumZ test will be performed. By default, all the traits will have the same weight.

If the user wishes to, she/he can specify a vector of weight by using the --custom-loadings option.

Access HDFStore components

Each table of the HDFStore is accessible through the command line tool jass extract-tsv (see command line reference for complete details).

jass extract-tsv --hdf5-table-path ./initTable.hdf5 --tsv-path './test_extract.tsv' --table-key SumStatTab

Alternately, you can use directly pandas read_hdf functions :

For instance if you want to access the Regions table :

pd.read_hdf("WK_test.hdf5", "Regions")

Note that is you wish that the SumStatTab table to be saved as a csv file you can provide the command lines with the --csv-file-path option and a csv will be generated as well. Outputting a csv while lengthen execution and require the appropriate storage space.

Command Line example

See command line usage for details .. code-block:: shell

jass create-project-data --init-table-path init_table/init_table_EUR_not_imputed.hdf5 --phenotype z_MAGIC_GLUCOSE-TOLERANCE z_MAGIC_FAST-GLUCOSE z_MAGIC_FAST-INSULIN z_MAGIC_HBA1C --worktable-path ./work_glycemic.hdf5 --manhattan-plot-path ./manhattan_glycemic.png --quadrant-plot-path ./quadrant_glycemic.png