Create a matrix of genetic correlations between a set of GWAS summary statistics
Current version: 2.2.0
- "munged" GWAS summary statistics, see https://github.com/bulik/ldsc
- Python version >=2.7 & <3; preferably 2.7.9
NOTE: by default, the script assumes that all munged sumstats are located in the working directory. This can be changed using the -i-flag (see below)
- 5x tab-delmited text files; N rows by (N+1) columns:
- rg.tab: symmetric matrix of genetic correlations
- rg_se.tab: symmetric matrix of SE's for the genetic correlations
- rg_p.tab: symmetric matrix of P-values for the genetic correlations
- cti.tab: symmetric matrix of cross-trait-intercepts
- cti_se.tab: symmetric matrix of SE's for the cross-trait-intercepts
- 1x tab-delimited text file containing all estimated values reported by LDSC; (N+1) rows (including header) by 12 columns
NOTE: by default, the output-files are saved in the working directory. This can be changed using the -n-flag (see below)
Run the script using the following command: bash make_rg_mat.sh
The script requires at least the three following flags:
Supply the location of ldsc.py (see https://github.com/bulik/ldsc)
-l [value]
Indicate which files to use as reference LD scores (see https://github.com/bulik/ldsc)
-r [value]
Indicate which files to use as regression weights (see https://github.com/bulik/ldsc)
-w [value]
Restrict the diagonal of both rg.tab and cti.tab to all 1's, restrict the diagonal of rg_se.tab and cti_se.tab to 0's, and restrict the diagonal of rg_p.tab to "NA"
-c
Save LDSC log-files
-s
Add prefix to output files
-n [value]
Change the number of parallel runs of LDSC to start; default=5
-p [value]
Input directory; defaults to working directory
-i [value]