Running Step 1 on the same dataset in v3.2.5 vs v3.2.5.3 results in the later version occasionally not converging with the following error message:
phenotype 5 (<MY_PHENO>) : Level 1 model did not converge. LOCO predictions calculations are skipped.
As I've said, all other params, data everything is the same, the only difference is the version of REGENIE, so my guess is that something went wrong with the v3.2.5.3 update. Shame, as I was interested in using it for my GxE project, as this new version was supposed to have a fix for that.
PS: I've also tried changing the niter, maxstep-null and maxiter-null parameters, but to no avail.
Running Step 1 on the same dataset in v3.2.5 vs v3.2.5.3 results in the later version occasionally not converging with the following error message:
phenotype 5 (<MY_PHENO>) : Level 1 model did not converge. LOCO predictions calculations are skipped.
As I've said, all other params, data everything is the same, the only difference is the version of REGENIE, so my guess is that something went wrong with the v3.2.5.3 update. Shame, as I was interested in using it for my GxE project, as this new version was supposed to have a fix for that.
PS: I've also tried changing the
niter
,maxstep-null
andmaxiter-null
parameters, but to no avail.