Error during ICA

[hanketim]

I want to use the script for MeDeCom with my own 76 samples. Up to the removal of the confounder by means of ICA everything worked without problems. I worked with DecompPipeline 1.0.3 and MeDeCom 1.0.1 This is the code I used:

data.prep <- prepare.data(rnb.set=data.prep$rnb.set.filtered,
normalization = "none", analysis.name = "Glioblastoma",
filter.beads = FALSE,
filter.intensity = FALSE,
remove.ICA = TRUE,
filter.na = FALSE,
filter.context = FALSE,
filter.snp = FALSE,
filter.sex.chromosomes = FALSE,
filter.cross.reactive = FALSE,
conf.fact.ICA = c("Geschlecht","AlterbeiED"),
ica.setting = c("alpha.fact"=1e-5,"save.report"=TRUE,
"ntry"=10,"nmin"=20,"nmax"=50,"ncores"=15))

however now the following error occurs:

Analyse stability
*** Starting parallel calculation on 10 core(s)...
*** System: unix
*** 50 components, 10 runs, 518989 features, 76 samples.
*** Start time: 2023-10-26 18:35:31.038152
*** Done!
*** End time: 2023-10-26 19:06:30.023182
Time difference of 30.98308 mins
Calculate ||X-SxM|| and r2 between component weights
Correlate rows of S between tries
Build consensus ICA
Analyse stability

Error in if (!fact %in% names(Var)) { : the condition has length > 1

For me it looks like 50 components are loaded (nmax = 50), but after that a problem occurs, which i don't understand. I would be very grateful for any help. If further information / details are needed, I will of course gladly provide them.

[schmic05]

Hi @hanketim,

Thanks for reporting this issue: it arose from R not allowing more than one condition in an if statement.
I fixed the issue in DecompPipeline with the most recent commit. Please try to update DecompPipeline and let us know if it works afterwards!

Best,

Michael

[hanketim]

Dear @schmic05 ,

thank you very much for your quick and helpful reply. That solved the problem and the confounders were removed! Now I can start with the methylome deconvolution, I am very excited!

Best regards,
Tim