Structural biologists use Macromolecular X-ray crystallography (MX), Nuclear Magnetic Resonance (NMR) and cryo Electron Microscopy (cryo-EM) to determine 3D structures of macromolecules such as proteins, DNA and RNA. To address complex research questions structural biologists is now stretching towards cellular length scales with techniques such as cryo-EM tomography and X-ray imaging while capturing dynamics via correlative microscopy and molecular dynamics (MD) simulations. The multi-purpose and multi-technique approach to characterize how macromolecules and their assemblies interact with virus and other pathogens in space and time is known as integrated structural biology. Today structural biologists experience an increasing amount of experimental raw data being collected by modern photon counting detectors at national and international facilities. Swedish structural biology research groups using the Swedish MaxIV light source for macromolecular X-ray crystallography and the cryo-EM facilities at Science For Life laboratories are generated large amounts of raw data that require supportive and research driven computational and storage solutions.
In november 2015, MaxIV and NSC formally decided to develop a HPC platform for MX in support of Swedish MaxIV users giving the PReSTO project the status of a MaxIV satellite. The MX installation made by NSC below is meant to be shared across Swedish HPC centers that want to participate in PReSTO and we therefore select to install all software at NSC Triolith using Easybuild for later move into with LUNARC Aurora. The Aurora installation of PReSTO is now available for academic use.
Swedish structural biologists can access the PReSTO project today by requesting membership in its predessesor pilot project (SNIC 2017/1-199) according to:
The pilot installation is a standard HPC installation, while the PReSTO installation has been performed by Easybuild for rapid sharing with other Swedish HPC centers coordinated by SNIC. The PReSTO installation also enable software launching from the Linux Desktop menu under PReSTO for my own Linux computer look and feel. Easybuild also enable transfer of environment variables to the compute nodes as been shown important to MX software that depend on Perl such as pipedream from GlobalPhasing.
Today at Aurora every user get 500 GB diskspace under /lunarc/nobackup/users/username while at Triolith we share 2.5 TB of diskspace under /proj/xray/users/username however if this is too small for your project or if you plan to run molecular dynamics simulations, apply for your own compute time at NSC Triolith or LUNARC Aurora. Pilot project membership grant access to MX software however having your own compute time allocation is required when combining MX with compute intensive molecular dynamics or Cryo-EM calculations.
This homepage contains startup instructions for the Easybuild installation of MX software that we call PReSTO. We attempt to guide HPC beginners in using the PReSTO environment and simply share links to excellent MX software guides and tutorials generated by MX software developers.
We are indepted to all MX software authors that kindly share their software for academic use in an HPC environment. We list all MX software currently in PReSTO installation with links to software home page and citations. Specifically we want to mention Gérhard Bricogne and Claus Flensburg from GlobalPhasing supporting the PReSTO project since its very beginning and shared all their MX software for academic HPC use including autoPROC with Staraniso for convenient elliptical scaling of diffraction datasets. We also want to mention the developers behind XDSAPP version 2.99 (Karine Röwer, Uwe Müller and Manfred Weiss) sharing a pre-release that can process Eiger data containers directly and Graeme Winter from the DIALS development team sharing a BioMax specific software patch for DIALS.
how to get started
explain basic HPC commands
BioMax data processing keywords
Triolith multi-node example
login, compute nodes and OpenGL acceleration
MX software home page and citations
forkxds enable multi-node XDS
convenient GUI launching at single compute/login node
sending parallel jobs to the queue
XDS, XDSAPP benchmark at Aurora vs. Triolith
Launch MX sw script at single compute node