Hi Tae-Hoon, The rlm is a function to solve robustly systems of equations. In ACNE, is used in the last step of the algorithm to scale the copy numbers of each of the alleles. The default objective function for rlm is "Huber" that in turn is a convex optimization optimization problem. So increasing the number of iterations would give better results.
However, we have not had any problems using the default values (in spite of the warnings). Usually the SNPs where rlm complains do not behave well and increasing the number of iterations is not worthwhile. Anyway, we will work on this and include the option to increase the number of iterations (now it does not have this option). Best regards, Angel On May 18, 5:48 am, Tae-Hoon Chung <hoontaech...@gmail.com> wrote: > Hi, All; > > It is about NmfSnpPlm() function. > When I ran it with some test data, it gave me a bunch of the same warnings: > > In rlm.default(t(H), matrix(data = 2, nrow = ncol(H), ... : > rlm failed to converge in 20 steps > > Since this is a warning, I know nothing tragic happened but, since some job > was not properly converged (I guess it has something to do with NMF), I > guess it casts doubts on the final results. > How serious is this non-convergence and is there any way to make the > procedure try harder (maybe longer iterations or so)? > > Thanks in advance, > TH > > -- > Tae-Hoon Chung, PhD > > -- > When reporting problems on aroma.affymetrix, make sure 1) to run the latest > version of the package, 2) to report the output of sessionInfo() and > traceback(), and 3) to post a complete code example. > > You received this message because you are subscribed to the Google Groups > "aroma.affymetrix" group with websitehttp://www.aroma-project.org/. > To post to this group, send email to aroma-affymetrix@googlegroups.com > To unsubscribe and other options, go tohttp://www.aroma-project.org/forum/ -- When reporting problems on aroma.affymetrix, make sure 1) to run the latest version of the package, 2) to report the output of sessionInfo() and traceback(), and 3) to post a complete code example. You received this message because you are subscribed to the Google Groups "aroma.affymetrix" group with website http://www.aroma-project.org/. To post to this group, send email to aroma-affymetrix@googlegroups.com To unsubscribe and other options, go to http://www.aroma-project.org/forum/
