Hi,

   My DESeq2 running was stucked for one night, it is normal?  My dataset contains 635 human samples, this is abnormally large:

head(ddsTxi)
class: DESeqDataSet
dim: 6 635
metadata(1): version
assays(2): counts avgTxLength

dds <- DESeq(ddsTxi)
estimating size factors
  Note: levels of factors in the design contain characters other than
  letters, numbers, '_' and '.'. It is recommended (but not required) to use
  only letters, numbers, and delimiters '_' or '.', as these are safe characters
  for column names in R. [This is a message, not an warning or error]
using 'avgTxLength' from assays(dds), correcting for library size
estimating dispersions
gene-wise dispersion estimates
mean-dispersion relationship
  Note: levels of factors in the design contain characters other than
  letters, numbers, '_' and '.'. It is recommended (but not required) to use
  only letters, numbers, and delimiters '_' or '.', as these are safe characters
  for column names in R. [This is a message, not an warning or error]
final dispersion estimates
  Note: levels of factors in the design contain characters other than
  letters, numbers, '_' and '.'. It is recommended (but not required) to use
  only letters, numbers, and delimiters '_' or '.', as these are safe characters
  for column names in R. [This is a message, not an warning or error]

Then it got stuck....Shall I wait or should I run it again step by step? or Can I just stop it and run the last nbinomWaldTest from the current dds?

dds <- estimateSizeFactors(dds)
dds <- estimateDispersions(dds)
dds <- nbinomWaldTest(dds)

Regards,
Raymond
 

What is your design? How many levels per variable?

Usually DESeq2 doesn't take so much time unless there are dozens of variables and hundreds of samples.

> dds <- makeExampleDESeqDataSet(n=100, m=600)
> system.time({ dds <- DESeq(dds, quiet=TRUE) })
   user  system elapsed
 37.552   4.407  42.106

It should scale linearly, so for 10,000 genes, you'd expect 70 minutes using a single core.

If you use parallel=TRUE, and 10 cores, this would take probably ~10 minutes.

You can filter out lowly expressed genes to save time, or switch to using limma-voom.

In my lab we use limma-voom whenever we have hundreds of samples.