Lowering tech barriers to make advanced computing accessible

High powered computers are a vital tool for scientists, researchers and academics who use software to simulate real-life situations. It can be used in the development of everything from medication to cars. However, it can be expensive and time consuming, requiring specialist skills to write submission files for each simulation (a new product could need 1,000 simulations) and working knowledge of different system interfaces.

A solution was needed in order to save time and money for researchers and businesses.

Enter, Longbow…

Longbow is a tool which runs simulations on our supercomputers, while delivering it to the users in ways which they are used to on their own laptops and computers.

This is what data looks like when it is benchmarked through Longbow
(Credit: HECBioSim/STFC)

For researchers and scientists, Longbow can:

mimic user software, which means that technical barriers previously associated with high powered computing (HPC) are removed
save time by automatically creating submission files which traditionally have been manually written for each simulation
save money which will enhance research grants and other funding
write and check submission files – ruling out errors before they can occur
transfer files seamlessly – input files from the user are sent into Longbow, and the results come back again to users – so no matter where in the world the simulations are being run, they can access it like any other program on their computer.

User friendly and accessible

Longbow is suitable for both novice users and experienced simulation experts. Those new to running simulations can be confident in the configurations that Longbow is creating, without having to undergo lengthy training. It also gives experts scope to run more simulations than would have otherwise been possible.

Longbow began as a collaboration between STFC and Nottingham University to produce a tool to simplify molecular dynamics simulations. It is now part of the High End Computing consortium for Biosimulation (HECBioSim) and STFC has a full time member of the computing community in the HECBioSim to provide high powered computing support for the biosimulation community. Additionally, biosimulation professionals can now also access some of the fastest GPU processors on JADE – a UK Tier 2 machine – via the consortium, with cases being screened by peers and with the consortia being on hand for support.

JADE is a UK Tier-2 resource, funded by EPSRC, owned by University of Oxford and hosted at the Hartree Centre. The hardware was supplied and integrated by ATOS Bull

It has already been used successfully in a number of experiments, including:

new drug discovery and design
development of more environmentally-friendly engines
greater understanding of DNA.

It is much more than just stand alone software. Longbow can be included in existing data processing software, such that if the compute load is light, an in-house system will analyse it. But as soon as it becomes too much for the system to handle, logic can be included that calls upon Longbow to offload it to much larger compute resources, where all file transfer and submission is seamless to the user.

Case study: computer-aided drug design

At Edinburgh University, Dr Julien Michel’s group uses Longbow in computational projects that feature a large number of protein and protein-ligand molecular dynamic simulations. A ligand is a small molecule that is able to bind proteins, and sometimes serves as a signal-triggering molecule.

One of these projects focuses on understanding loop dynamics – how system process variables behave in response to sudden changes – in a family of proteins called Cyclophilins. These are potential drug targets for a number of conditions, such as Hepatitis C, HIV, cancers and neurodegenerative disorders. Clinical validation requires the development of better and more selective inhibitors so Dr Michel’s group is running simulations to help advance this development.

“We are exploring the use of Longbow in prototype software we are currently developing. One project focuses on a virtual-reality user interface for interactive MD simulations. The other project focuses on an adaptive protein MD simulations workflow to discover cryptic pockets in protein surfaces. These pockets may be used to formulate new strategies to discover drugs for treating a wide range of diseases”.

Dr Julien Michel

Longbow’s successes include new drug development
(Credit: Pixabay)

Not standing still

2018 saw improvements to ensure it continues to be resilient to technology changes. The coming year will see more support for its use beyond biology for example in chemistry, physics or engineering.

Work is currently ongoing to deliver a state of the art training portal based on a cloud platform which will give us the capability to deliver all of our community training needs in a web browser via the open-source Jupyter notebooks. They are also benchmarking codes that are commonly used within the biosimulation community to show how they perform on HPC machines around the country. This allows academics to appropriately set their level of resource to ask for when applying to use these big machines.

Longbow is just one piece of work detailed in the latest Science Highlights report – published by STFC’s Scientific Computing Department. You can download Science Highlights 2018 here.

Last updated: 12 April 2019