How can I run GOLD in parallel?

For very large virtual-screening tasks, we provide the GOLD Cluster tool. For smaller-scale parallelization, e.g. on a computational chemist's workstation or a local compute server, we do not have a specific product, but options are available. 

Through one of our Integration Partners

GOLD can be run in parallel via one of the integrations we have with various third party cloud-based systems:

• OpenEye's Orion
• CCG's MOE
• Biovia's Discovery Studio

KNIME

CCDC Provides an integration to KNIME via our CSD Python API that allows you to use GOLD fuctionality.

GOLD Cluster

This tool runs GOLD, inside a Singularity container, on an HPC compute cluster managed by Slurm. The GOLD Cluster tool and its User Guide are available to customers via our Downloads site under 'CSD-Discovery > GOLD'. 

Smaller-scale parallelization

GOLD Cluster can be run on a Slurm cluster running in a Linux VM. Thus, in principle, this tool could be used to parallelize GOLD docking on a workstation. However, the setup involved isn't trivial and might not be straightforward on all platforms. A simpler solution for this use-case is thus required and a script is available which illustrates the use the CSD Docking API and the Python standard-library multiprocessing module to parallelize GOLD docking. This approach should be suitable for docking some hundreds or perhaps thousands of ligands, depending on the compute resource available and the docking protocol chosen. For further information, please contact support@ccdc.cam.ac.uk.

CSD Python API scripts repository (the script mentioned above is in gold_multi)

Custom Solution

Many of our customers utilise GOLD in parallel using their own custom solution for parallelisation of our standard GOLD software package. While not fully supported, this does allow users the flexibility to integrate GOLD with their existing workflows.

Platforms no longer supported

We no longer support the GOLD Cloud tool (which ran GOLD inside a Docker container on a Kubernetes cluster) or the use of GOLD with Parallel Virtual Machine (PVM). We also do not currently provide tooling or support for running GOLD on Grid Engines, although there is nothing in principle that should prevent this being done.