FAQs

Can I extract or upload ONT data to the GDC home or GDC projects?

No, raw ONT data contains of millions of small files. You must keep them as archives (tar) on GDC home or GDC projects and only extract them in the scratch. For more information please contact us.

Should I unzip fastq files?

Never unzip files in GDC home or GDC projects many tools can handle them. If you still like to uncompress them you need to use the scratch.

Why can't I access Euler?

Make sure that you are connected with the VPN.

Can I run jobs over the weekends?

We do not provide support out of hours, so it is your responsibility to ensure that jobs are running properly. For less advanced users we recommend working during office hours.

When submitting my script to slurm I get strange error massages and no job has been submitted. What is the problem?

Is the submission header complete including quotes? Make also sure that you have linux line breaks.

mac2unix file
dos2unix file

Do I have to constantly monitore my jobs?

New scripts need to be constantly monitored and adjusted in case the usage (CPU, memory) is too low. Established scripts needs to be monitored from time to time. If you run jobs on Euler you need to be available via email and react soon.

How can I reset my passwords?

There are two different passwords:
(1) LADP: Euler, Lab server
(2) Radius: VPN
If you like to change the passwords check out this video.

Should I split my fasta files in junks if I like to use tools like Kaiju, Kraken or Diamond?

No these tools have been developed to process large fasta or fastq files. Instead of a conventional blast search the database will be loaded in the memory to speed up the searches massively.

How much memory do I need for a mapping?

Mapping is normally CPU intensiv. The amount of memory is dependant on the size of your genome. For bwa mem you need 16 Gb for a human genome, whereas for bwa-mem2 you need a around 25G. The scheduler often report more memory (cache memmory) than what you need physically.

I don't have the experience to work in an HPC environment. Do I need to monitore my jobs?

Sure but monitoring is not difficult, it just takes some time. Use the tools available. We can also help you set up scripts.

How can I monitore my jobs?

running jobs: myjobs
finished jobs: Slurm job webGUI
summary jobs: get_inefficient_jobs

I want to install tools in a conda environment?

Don't use conda on a HPC. For Python tools, use virtual environments. For complex tools, container solutions (singularity, dockers) are better. We can help you set this up.

Why do my Java jobs always get killed?

For Java tools (GATK, picard, beagle, bbmap) make sure you define the environmental variables correctly. Only incease memory if you get a heap space error.

I can't find a tool in the software stack?

Make sure you have the GDCstack sourced and use module --show-hidden avail. If you are missing an important tool, let us know and we will see if we can install it for you.