Working with data

Where to put your data

On your laptop, all files reside on your local hard disk. However, on GenomeDK that is not the case. To achieve high performance and accommodate the huge amounts of data located on the cluster, data is saved on network file systems.

Each of these file systems have their own advantages and disadvantages. Thus, you need to put your data on the right file system to utilize the cluster optimally.

Storage locations


Accessible by


Eligible for backup

Has snapshots



Only you




100 GB


Only you





/faststorage/project/<project name>

Project members






Only you






Your home folder is on a file system called NFS. This file system handles many small files well, but not big files. It’s also rather slow to access this file system from many compute nodes at the same time and your home folder is limited to 100 GB.

We do not recommend storing raw data files, temporary files or results on this file system. However, it’s fine for small documents like notes, programs and your Conda installation.

This location IS NOT eligible for backup. Instead we perform regular snapshots (see Snapshots).


For big files we provide access to a fast, parallel file system called BeeGFS. We call this file system faststorage. This file system is ideal for large data files and intermediate data generated by your jobs. However, it will be much slower for small files. All users have their own home folder located at /faststorage/home/<username>. There is no limit on the amount of data that can be stored in this folder, but the data can only be accessed by your user.

This location IS eligible for backup.

/faststorage/project/<project name>

All projects get their own folder on fast storage. All files related to the project should be placed in this folder. Project folders have no quota and can be accessed by all members of the project.

This location IS eligible for backup.


All jobs get access shared, local disk. Files written to this space are only available on the given node (that is, you can not access files written here from other nodes).

This folder is deleted when the job finishes! Any files written here, that you wish to keep, should be moved to faststorage as the last step in your job script.

Use this for (1) temporary files that are not needed later, or (2) for output files that are written in small chunks.

This location is NOT eligible for backup.

Backing up data

We provide backup on good old-fashioned tape to all users. To back up a file, it should be put in a directory called either BACKUP, Backup or backup. The directory can be located in any other directory, but only some locations are eligible for backup (see Where to put your data).

Data is backed up approximately once per week and is kept for 30 days.


Do not back up temporary data files that can easily be reproduced. Computation is cheap, but backup is very expensive. The backup is meant for scripts/source code and important raw data.


All home folders are automatically snapshotted. As a user you do not have to do anything to enable this.

Snapshots do not provide the same data safety as a backup since the data is not stored at an offsite location. Instead, a snapshot of the data is taken on a regular basis and stored on the local server. This means that if you e.g. deleted a file by mistake, the file can be recovered from a previous snapshot.

Snapshots are currently taken with the following intervals and retention times:

  • 15 minutes, last 4 are kept

  • 1 hour, last 24 are kept

  • 24 hours, last 31 are kept

  • Weekly, last 8 are kept

  • Monthly, last 12 are kept

Accessing your files locally

You can access your files on GenomeDK locally by a process called mounting. Mounting the GenomeDK filesystem locally makes it possible to access and edit your files as if they were located in a folder on your own harddrive.

Unfortunately, mounting over SSH does not work on Windows. If you’re on Windows you can use MobaXterm or one of the alternatives listed in Copying data.

  • On distros with the apt package manager (Ubuntu, Mint etc.):

    [local]$ apt-get install sshfs
  • On distros with the yum package manager (Fedora, CentOS etc.):

    [local]$ yum install sshfs
  • On macOS, download and install the SSHFS and FUSE for macOS packages from the OSX FUSE website.

Create a directory where the filesystem will be mounted:

[local]$ mkdir ~/GenomeDK

Now mount the filesystem by running this command:

[local]$ sshfs ~/GenomeDK \
    -o idmap=none -o uid=$(id -u),gid=$(id -g) \
    -o allow_other -o umask=077 -o follow_symlinks

Where USERNAME should be replaced with your GenomeDK username. You should now be able to access your files on GenomeDK by going to the ~/GenomeDK directory on your computer.

To unmount the directory, run:

[local]$ umount ~/GenomeDK

Copying data

Closed zone…

To copy data from the cluster, see Exporting files To copy data to the cluster, see Importing files

From your own machine to/from the cluster

If you mounted GenomeDK on your computer, you can copy files to and from the cluster by simple drag-and-drop. Otherwise you can use one of the solutions listed here or one of these alternatives:

You may also use the command line.

To copy a single file from your computer to the cluster:

[local]$ scp myfile.txt

On Windows, replace scp with scp.exe.

To copy a single file from the cluster to your computer:

[local]$ scp .

If you want to copy an entire folder to/from the cluster you will want to use rsync instead. To copy a folder from your computer to the cluster:

[local]$ rsync -e ssh -avz /path/to/data

Windows doesn’t have rsync installed, so you must resort to one of the options listed above.

If you want to upload a folder, but also delete files that you deleted in the source folder from the destination:

[local]$ rsync -e ssh -avz --delete /path/to/data

If you want to download data from the cluster:

[local]$ rsync -e ssh -avz --delete /location/data

You may want to add the --progress flag to all of these commands if you’re downloading/uploading large amounts of data.

From the Internet to the cluster

You can use wget to download data from the Internet to the cluster:

[fe1]$ wget -c --timeout=120 --waitretry=60 \
    --tries=10000 --retry-connrefused URL

Remember to replace URL with the thing you want to download.

When downloading large files you are encouraged to limit the progress output to avoid stressing the system, especially when you’re sending the progress output to a file:

[fe1]$ wget -c --progress=dot:giga --timeout=120 --waitretry=60 \
    --tries=10000 --retry-connrefused URL

Editing files

If you mounted GenomeDK on your computer, you can edit files directly by just opening them with your prefered text editor on your computer. Otherwise you can use one of the solutions listed here.

Nano, vim, emacs

With editors like nano, vim and emacs you can edit files directly on the cluster. The editor itself also runs on the cluster and thus your editor settings etc. are conserved, even if you log in from another computer. Also, these editors don’t require a graphical user interface, so you don’t need X-forwarding or VNC.

The nano editor is by far the simplest editor of three, but also the least powerful. However, it’s just fine for quickly editing scripts or looking at output files. The documentation for nano can be reached by running the command:

[fe1]$ man nano

You can open nano by running:

[fe1]$ nano name-of-file.txt

Likewise, vim and emacs are already installed on the cluster. Documentation for each editor can be found on their respective websites.

Gedit with X-forwarding

If you want a graphical user interface and a more familiar editing experience, you may use the Gedit editor with X-forwarding. Make sure that you are connected to the cluster with X-forwarding enabled. Then run:

[fe1]$ gedit

This will open the Gedit editor in a new window. Since the editor runs on the frontend, you have access to all of your files on the cluster.

Encrypting sensitive data

If you need to transfer sensitive data (for example human genomes) out of the cluster you must encrypt the data first. Encrypting the data makes it impossible for strangers to look at it without decrypting it, which requires a password chosen by you.


[fe1]$ openssl aes-256-cbc -a -salt -in data.txt -out data.txt.enc

This will encrypt data.txt and write the encrypted data to data.txt.enc. You will be prompted for a password which is needed to decrypt the file again.


[fe1]$ openssl aes-256-cbc -d -a -in data.txt.enc -out

This will ask for the password used to encrypt the file. The decrypted contents are written to

Prevent accidental changes to data

Put the data in a separate folder and run:

[fe1]$ chmod -R a-w <folder name>

Now you can’t change, add or remove files in that folder or any of its subfolders.