RPMM blc() function not completing?

Question

[mjt.professional]

Hello,

I don't see many questions on the forum regarding the RPMM package. In particular, I am using the BMIQ script/method to normalize the Type 2 probes on the Illumina 450K DNA methylation chip. This procedure uses a piece-wise beta distribution fitting procedure to then recalculate methylation values for these probes.

The BMIQ script calls the blc() function from the RPMM package. I've noticed that fairly often, this function simply hangs. It has both a maximum iterations parameter as well as a convergence tolerance threshold. Neither of these seem to matter in terms of this bad behavior of the function.

Has anyone encountered this problem?

Thanks

Answer 1

[andrew.j.skelton73]

Have you considered trying the wateRmelon implementation of BMIQ, it might be more stable than RPMM's

[mjt.professional]

I'll give that a shot. This is rather maddening as it's hanging somewhere deep in the code dependencies and seems to depend on the random seeding of the core optimization routine.