This job is waiting to run

12 months ago by

United States

Jennifer Hillman Jackson ♦ 25k wrote:

Hello,

The RNASTAR jobs started less than 14 hours ago are still queued and the other mapping jobs started a few hours ago (Bowtie2/BWA-MEM) are currently executing. These likely were sent to different clusters. There are no known issues right now, but we are double checking and will post an update if a problem is discovered.

I did notice that the custom reference genome is not in fasta format. It is instead a count of the lines of the original fasta dataset. This means that all jobs will end up failing, both those that are queued and those that are executing. All will need to be permanently deleted and rerun using a valid fasta formatted custom reference genome.

FAQ: https://galaxyproject.org/learn/custom-genomes/

Thanks! Jen, Galaxy team

ADD COMMENT • link written 12 months ago by Jennifer Hillman Jackson ♦ 25k

Update: We confirmed that the queued RNASTAR jobs were sent to a cluster that happens to be very busy. If these had used a proper fasta custom genome, leaving them to run would be best. But for your case, since the input has a problem, rerunning with the corrected input is needed.

I ran a set of small test jobs against all clusters and these have completed now except for the one queued at PSC Bridges. This is the busiest cluster right now. Which clusters at any time are busy changes over time and when using default Job resources, any of the clusters can be picked (a process that sometimes uses filters to match up predicted resource needs). When you rerun your jobs, you could consider using the "Specify Job Resource Parameters" option near the bottom of the tool form and select ether Jetstream or the default (local) cluster directly.

The jobs might run faster - but this is true just right now - and you don't need to avoid the PCS Bridges cluster ongoing. If that changes or a problem with the cluster is uncovered, we'll write back with another update.

Just so you know, RNASTAR requires much more memory than Bowtie2/BWA. It also expects spliced RNA fastq input the same as HISAT2 (which in many cases is a better tool choice - it also uses much less memory during job execution). Bowtie2/BWA expect unspliced DNA fastq input when mapping against most full genomes but for some use cases would be fine to use when mapping RNA against a species/genome that does not contain introns (example: most bacterial genomes) or a transcriptome/exome.

My guess is that the DNA mapping tools were a test, yet this seemed worth clarifying. Please see the underlying tool/s help and manuals to fully understand the parameters and expected inputs.

Thanks for your patience! Jen

ADD REPLY • link modified 12 months ago • written 12 months ago by Jennifer Hillman Jackson ♦ 25k

Similar posts • Search »