Hi,
I have been trying to utilize apeglm
for shrinking , and I am receiving the error message below; however, normal
works just fine...I am using DESeq2_1.18.1
and apeglm_1.0.3.
resLFC <- lfcShrink(dds, coef="condition_treated_vs_untreated", type="apeglm")
using 'apeglm' for LFC shrinkage
Error in solve.default(o$hessian) :
system is computationally singular: reciprocal condition number = 1.36798e-116
Hi Michael,
Here are the dimension of the model:
type = "ashr"
gives me this type of warning message:I have tried to install the
REBayes
package, but I am currently having issues installing the package because it is dependent onRmosek
which requiresmosek
on the local machine and themosek
installation is giving me issues.....Update: I successfully installed all of the dependencies, and now
type = "ashr"
is working.Sorry just got to this. The message above is not an error. You can also use ashr with the EM algorithm which happens if it can’t find this other package (although I suppose the REBayes is preferred and we have installed this on our cluster).
OK. I successfully installed
REBayes
, but users should know that it has dependencies toRmosek
, which requires instillation ofmosek
to one's local computer, which requires a license... so it was a bit of a long process to install theREBayes
package.Thanks!