No variables selected

[wiedenhoeft]

When running BMSel on a large dataset of ~17,000 markers, both lasso and lasso-pcvl fail to select any variable for the full dataset, though they do select some for subsets of a few hundred markers. For a couple of markers, I get "(only one value available for all observations)", and for about 5 lambdas (in the 80s and 90s), I get "from glmnet Fortran code (error code -93); Convergence for 93th lambda value not reached after maxit=100000 iterations; solutions for larger lambdas returned".

1. Is there a way to increase maxit through the biospear interface.

2. Failing to select even a single variable seems like an impossibility to me, could this be a bug?

Thanks,
John

[wiedenhoeft]

P.S.: Previous to using biospear, I have used a modification of the code in the paper supplement. In case this would be helpful, I'm attaching the source here, since it does actually select variables.
pcvl.R.txt