lm1 in the C++ was not being initialized. This has been fixed.bama now contains fdr.bama, a bama variant that controls for the false discovery rate by estimating the null PIP distribution via the permutation test. It contains options to run in parallel (ie using the parallel package) as well, since the permutation test can be time consuming. fdr.bama also contains a summary function, summary.fdr.bama to help summarize results.bama (and fdr.bama) now contain a weights parameter, weights.bama documentation.bama type checking.The most notable feature in this release is the renaming of hdbm to bama to prevent confusion with regards to its relationship with hdmm.
bama now only accepts a single extra covariate matrix C, instead of C1 and C2. ## Minor Changes
Added a summary function to bama to calculate the PIP, estimate, and credible interval of each mediator and optionally rank them.
Added data quality check to ensure the matrix column norms in M, C1 and C2 are not too small. If they are too small, NaNs could be generated because of overflow problems when dividing by the norm.
Improved documentation
Added checks to C, the matrix of extra covariates, and alpha.a, the initial alpha.a values.