Why does the GATE pipeline fail with no obvious reason
pipeline failure with convergence error related to covariates
In case your GATE pipeline fails with an error message similar to "missing value where TRUE/FALSE needed" the job has crashed while executing an R script. You can confirm this by checking the messages surrounding the error. They should contain hints to "package"s and may even be linked to a particular R package called Rcpp.
While this error cannot be sufficiently explained it can be avoided by reducing and/or changing your covariates, in particular the principal components (PCs) in your phenotype-covariate file. This trial-and-error approach is admittedly somewhat cumbersome but has proved successful as removing a small number of PCs prevents convergence from failing but should not dramatically affect the statistics.
When editing your phenotype-covariate file do not forget to modify your json file accordingly, changing the value informing on the used covariates.
Last updated
Was this helpful?