Skip to Content

11. Data processing

Version 28.1 by tabermah on 2024-03-21 15:14

11.1. Automatic data processing

Unless you un-check the option for the specific acquisition run, autoprocessing will be executed. For standard rotational data autoprocessing is via XDSAPP (https://www.helmholtz-berlin.de/forschung/oe/ps/macromolecular-crystallography/hzb-mx-software/xdsapp3/index_en.html, current version 3.1.9) (Sparta et al. (2016) J. Apple. Cryst. 49, 1085-1092) for strategy calculation we use Mosflm (Leslie Acta D62 (2006) 48-57) and Spotfinder (Zhang et al. (2006) J. Appl. Cryst. 39, 112–119) for producing the heatmap for the grid scans. The results are compiled into HTML-files to give you a convenient way of checking them. The output files are in the processed folders of the corresponding datasets.  A self refreshing index of all results is created in the beamtimes processed folder. During the beamtime you can open it from file::///gpfs/current/processed/index.html in a browser. The HTML-file is also contained in your data backup.

11.2. Manual data processing

Please do not use the control computers of the beamline (haspp11user01-04) nor the haspp11eval01 for data processing (latter is now fully dedicated to autoprocessing).

Use Maxwell-cluster for data processing as described in the next chapter. Access to Maxwell requires a Scientific Account. 

Manual data processing after the beamtime or during a remote session is only available for registered DESY Scientific Accounts.

Prerequisites

Local users can perform manual data processing as described at the beamline. Users having remote access can also process their data only if they are through their DESY Science Accounts. Users may apply for science accounts to analyze data measured at Petra III or Flash on resources located at the DESY computer center. Science accounts have a standard lifetime of 3 years (renewable).

Applying for a Science Account:

  1. the user submits a request to the beamline scientist/manager
  2. the user receives a pre-filled form by Email within a few business days from the DESY-FS Administrator
  3. the form has to be signed and sent back to the Administrator (hard-copy as scan/photo by email/fax in pdf format)
  4. the user receives the initial credentials from the beamline scientist/manager who forwarded the account request. If on site at DESY, the user can authenticate him-/herself by valid ID card or passport at FS-EC's Administrators to retrieve the initial password for the account.

Multi Factor Authentication (MFA) at DESY:

For accessing the DESY computers a two-factor authentication is needed. After you have received your Science Account, you will need to setup an Authenticator App to your smartphone. You can find the instructions here.

Accessing Data

There are three methods to access DESY computers from outside using a scientific account:

  1. Command line:
    in a Linux terminal or Cygwin (windows) type:
  2. ​Graphical User Interface:
    in a web browser type:
    • https://max-fs-display.desy.de:3389/
    • enter your Science Account credentials
    • enter your 2FA token
    • launch e.g. an XFCE session
    • open a terminal
    • cd /asap3/petra3/gpfs/p11/YEAR/data/YourBeamTimeID
  3. Installing and using the FastX3 client:​
    • visit Downloads DESYcloud. Login with your DESY account.
    • click on the desycloud link next to the starnet entry. If you get to an empty page, just reload the page.
    • go to StarNet_FastX3/client/3.3 (or the highest version listed) and download the package matching your OS
      1. FastX-*-setup.exe and FastX3.msi are regular windows installer
      2. FastX-*-setup_nonroot.exe is a "portable" windows installer, e.g. can be placed on a USB stick
      3. FastX3-*.dmg is the MacOSX installer
      4. FastX3-*.rhel7.x86_64.tar.gz is a tarball for Linux, works also for Ubuntu or Debian (possibly requiring some additional packages)
    • install and launch FastX3
    • configure a new https connection (host: max-fs-display.desy.de, port:3389, auth: SSH)
    • double click on the new connection
    • enter your 2FA token
    • launch e.g. an XFCE session
    • open a terminal
    • cd /asap3/petra3/gpfs/p11/YEAR/data/YourBeamTimeID

Pros & cons:​

  • ssh: simple, convenient, but no GPU acceleration and X11 over ssh is slow. Sessions are not persistent.
  • browser: simple, fast,works everywhere, GPU hardware acceleration, but has some limitations (copy&paste, browser capturing key shortcuts). Supports session sharing
  • client: fast, works best, but requires installation of a client
  • browser and client offer persistent sessions: when disconnected your session continues to run without any client connected. Reconnecting to the display server gives access to the running session.

Data is accessible to users who were registered participants to the experiment in Door prior to the beamtime ID being opened at the beamline. If you cannot access the data, please contact your LC or the beamline manager.

Data processing

The processing folders contain two folders, 'xdsapp' and 'manual'. 'Manual' contains a template for manual processing having all the correct parameters. However, to shorten the processing time and make use of the computational structure, reprocessing your data on the maxwell nodes should be done by using a script to queue the jobs in slurm:

  • a template script (xds.sh) can be found at /asap3/petra3/gpfs/common/p11/
  • the partition in the script is defined psx for external users (ps should be used for internal users of Photon Science)
  • copy the script to the folder where you run the processing
  • launch the script by typing sbatch xds.sh
  • the script will find a free node for your job to run fast
  • the xds-log you get in a file xds-job-###.out.

Processing Serial Crystallography data by nxds should be done via a script to queue the jobs in slurm:

  • a template script (nxds.sh) as well as an example of input file (nXDS.INP) can be found at /asap3/gpfs/common/p11/
  • the partition in the script is defined psx for external users (ps should be used for internal users of Photon Science)
  • copy the script and the nXDS.INP to the folder where you run the processing
  • the nXDS.INP file needs some editing to fit your data path and data collection parameters (wavelength, detector distance, oscillation...)
  • the processing is launch by typing sbatch nxds.sh

If manual processing through a terminal or GUI is absolutely necessary:

  • allocate a node to yourself by typing in terminal: salloc --node=1 --partition=psx --time=xx:xx:xx
  • ssh to the node allocated to you
  • load the modules of the needed crystallographic softwares on the node (instructions below)
  • HOX! terminate the allocation after usage by typing exit twice on your terminal window

Crystallographic and Scientific Software Packages

Software packages might be available out-of-the box as system installs or as part from the OS default repository, others are available via module functionality. A description on how to enable/use a particular software package is given on confluence.

Crystallographic software packages are in the Photon Science section.

You can store analysis results in scratch_cc (temporary/testing results) or processed (final results).

To use crystallographic programs type in terminal (for example):

  • module avail #to check what programs and versions are available
  • module load xray # sets the environment for xds and helper GUIs such as XDSAPP
  • module load ccp4/7 # sets the environment for ccp4 version 7
  • module load phenix # sets the environment for phenix

Once the correct module is loaded, the program can be called via the command line by typing its shortcut or alias (xdsapp, phenix, albula....etc).