rnaseq
rnaseq copied to clipboard
DESEQ2_QC_STAR_SALMON timeout with rlog and 400 samples
I ran rnaseq v3.8.1 on 400 samples. Failed as I got a timeout after 8hrs on DESEQ2_QC_STAR_SALMON process. It was still running rlog after 8 hours with the message:
rlog() may take a long time with 50 or more samples,
vst() is a much faster transformation
I have specified --deseq2_vst which works but i wondered if --deseq2_vst could be made default?
I second that it should be the default
Also, the timeout seems unnecessary - why produce an error when things could still finish correctly?
I also agree with this proposal. I have been processing hundreds of RNA-seq datasets and this step of the pipeline fails after 8 hours.
Furthermore, since I have set automatic retries of [attempt number] * 8 hours - this would cause the pipeline to fail again after 16 hours and then again after 24 hours.
Consequently, the rnaseq pipeline terminates in error and takes around 2 days to process this PCA plot, which is only for QC purposes.
I think Oliver’s suggestion of having --deseq2_vst as the default option is a very good idea and should improve the pipeline substantially when processing a large number of samples.
Otherwise, thanks for maintaining this great pipeline!
Best,
Steven
Agree too! Will be in the next release.
Fixed in https://github.com/nf-core/rnaseq/pull/874/commits/a4be75f4cc55c19ae8a1df34cfa8acc7afc0853e