Biologists looking at the structure of molecules like proteins and nucleic acids are interested in how they acquire their structures and how alterations may affect their function. This understanding is a key factor in modern disease treatment and prevention, and computational power such as that of the Open Science Grid (OSG) has become critical in advancing the field of structural biology.
Image courtesy Gunnar F. Schröder, used with permission
In 2000, Dr. Piotr Sliz established SBGrid to provide software and computing to fellow biologists. SBGrid has since expanded to support biology laboratories with a large library of scientific applications. It now has 245 member labs worldwide and supports over 270 scientific software applications. In 2010, SBGrid established a Virtual Organization (SBGrid VO) within the OSG, adding a key grid computing element. SBGrid soon became one of OSG’s biggest users and key partners. (Dr. Sliz is a member of the OSG Council, and SBGrid hosted the 2011 OSG All-Hands Meeting at Harvard Medical School.)
Mick Timony maintains and develops the SBGrid Science Portal—its high performance computing portal—and related OSG infrastructure. He also coordinates software installations for member labs and maintains the SBGrid website. Because Timony is a computer scientist, he can provide the kind of technical support that his non-technical scientists need for effective research.
Mick Timony, photo courtesy Brian Sherman
“The SBGrid started by building software for structural biologists at Harvard Medical School,” said Timony. “That effort grew into us becoming a bigger organization that built more and more software. There is a lot of software out there that is not easy to use. We take care of that for them. We use a chargeback model for many of our services. We also offer core computational services for local research groups, and anything we offer on the OSG is free to the community.”
Scientists used to be confined to big workstations. Now, says Timony, they want to run things on their desktop computer—and maybe at home, on a bus, or wherever is convenient for them. “They can’t always get a big workstation, so we create a laptop setup that allows them to select software to install. We set up an environment like they would have if they were physically in the lab.”
On the OSG side, SBGrid offers two workflows through the SBGrid Science Portal for free use by structural biologists worldwide. Wide Search Molecular Replacement (WSMR) performs an exhaustive molecular replacement search of the Protein Data Bank in a matter of a few hours. Such searches were once nearly impossible. The other workflow, Deformable Elastic Network (DEN), refines structures from low-resolution X-ray crystallographic data sets. This would take many thousands of hours on standard computational resources, but can be completed in several hours using the DEN portal and the OSG.
During 2015 the SBGid VO used nearly 3.4M CPU hours from shared OSG resources and in this time completed over one million structural biology simulations.
“The workflows are not complicated,” said Timony, “but running in parallel is key. That enables mass scale computing without time limitations. I can’t say enough about the great support we get from the OSG.”
“Scientific discovery is getting into the pipeline much more quickly, thanks to resources like the OSG,” concluded Timony. “The OSG is saving researchers lots of time and saving them from dead ends. They are able to identify results more quickly—what the protein is and what it might be used for. Scientists want to be in the lab. They don’t want to mess with software. They want to get their work done and the OSG is helping us to help them do just that.”
– Greg Moore