Back to index.html.

SNP solutions from GBLUP

Model

We can back-solve the SNP effects from the GBLUP solutions for breeding values, that is, direct genomic values (DGVs). A simple formula for obtaining the SNP effects is based on best linear prediction (BLP) and, equivalently, a generalized selection-index formula. See the textbook and Wang et al. (2012) for details.

The author of the textbook calculates the SNP effects for the data used in the previous section. We will perform the same computation using POSTGSF90, an additional tool for computing SNP effects from the solutions produced by BLUPF90+. Here, we demonstrate how to perform a “post-GBLUP” analysis using the BLUPF90 programs.

Files

The GitHub repository provides the files (mr114.dat, etc.) required for this analysis. These files are identical to those used in the previous section (mr113.dat, etc.). The procedure for predicting SNP effects is as follows.

1. Run BLUPF90+ (blupf90+) and obtain the breeding-value solutions in solutions.
2. Run POSTGSF90 (postGSf90) with the same parameter file.

Note that POSTGSF90 does not accept some options that are effective for BLUPF90+. For example, POSTGSF90 will stop if the parameter file contains OPTION solv_method FSPAK. You must remove such options before running POSTGSF90.

Solutions

Once POSTGSF90 finishes successfully, it generates several files. The SNP-effect solutions are available in snp_sol.

trait effect snp chr pos snp_effect weight variance_explained var_a_hat
         1         2         1         1         1  0.8691067E-01  0.4111146                  0                    0
         1         2         2         1         1 -0.3103946       3.526523                  0                    0
         1         2         3         1         1  0.2621196       3.936336                  0                    0
         1         2         4         1         1 -0.8018383E-01  0.3683551                  0                    0
         1         2         5         1         1  0.1100588      0.3701193                  0                    0
         1         2         6         1         1  0.1388949       1.148245                  0                    0
         1         2         7         1         1 -0.9949045E-09  0.1638275E-16              0                    0
         1         2         8         1         1 -0.3855329E-08  0.1277342E-15              0                    0
         1         2         9         1         1 -0.6060414E-01  0.2244535                  0                    0
         1         2        10         1         1 -0.1579683E-01  0.1485259E-01              0                    0

These values are identical to those in the textbook (\(\hat{\mathbf{g}}\); p. 195). We do not discuss this file or the other files generated by POSTGSF90 here. See the later chapter or the manual for details.

Back to index.html.