I am trying to replicate the workflow from library(DirichletMultinomial)
. I have large matrix containing counts count
and vector of factors pheno
. I made a subset of count
into countp
based on pheno
(division into groups) as in the example in vignettes shows countp <- count[pheno %in% c("group1", "group2"), ]
after executing the next step for Dirichlet-multinominal model for each group:
bestgrp <- dmngroup(countp, pheno, k=1:15, simplify = FALSE, verbose=TRUE, .lapply = parallel::mclapply)
the process was working for 8-12 hours and crashed reporting:
invalid class “DMNGroup” object: undefined class for slot "elementMetadata" ("DataTable_OR_NULL")
I know for sure that the process took several hours since crashed because I have following code
if(exists('bestgrp')) {
save(bestgrp, file = "NRR_DMM_bestgrp.Rda")
} else {
write(geterrmessage(), file = "!!!ERROR.txt") }
and have the time the error message was saved. Since verbose=TRUE
doesn't work in parallel multiple core computing I don't have more specific point when it crashed. The whole dataset count
when processed with dmn
goes smooth.
I am not a bioinformatician, and new in R computing therefore as Google returns NULL results for the error message I am totally lost.
Can anyone help?
Best regards, Marcin
question from:
https://stackoverflow.com/questions/65936924/what-does-the-error-invalid-class-dmngroup-object-mean-how-to-resolve-the-p 与恶龙缠斗过久,自身亦成为恶龙;凝视深渊过久,深渊将回以凝视…