Application of e-infrastructure to real research

Application of e-infrastructure to real research

Zhongwei Guan, EngineeringModelling engineering structures with complex features in terms of materials, geometries, loading conditions, interactions, etc.

SLM lattice structure

PVC sandwich under blast FMLs under blast

FMLs under impact

What we did:

• Purchased CPU hours from STFC after trying it on NGS Benefits:• Accessibility: anywhere in the world as long as there is a high speed internet connection• Updated version of commercial packages• More frequent updating the supercomputer than a local network • It is a low cost service (10000 CPU hours/£1000)

What we found using e-infrastructure:

• High efficiency, saved a lot of time for me, can submit a few jobs at the same time • Big jobs which cannot be run locally• large disk quota (50 GB or more) and CPU allocation with low costs• Regular backup• Reliable and quick help from NGS staff• PhD and postdoc projects• Available commercial packages, Abaqus, Fluent, Fortran, etc.

Rebecca Notman, biomoleculesMolecular dynamics simulations of biomolecules

and biomaterials

How do nanoparticles get into cells?

Can we use nanofibres to deliver drugs to the

brain?

Can we use biomolecules to synthesize and assemble complex nanomaterials?

Use of e-Infrastructure

• Use different types of e-Infrastructure depending on the types of calculation we want to do

Local resources:– Free energy calculations of amino acids binding to

quartz (small model system, runs in serial) National grid service:

– Replica exchange molecular dynamics to explore peptide conformation

– Many calculations running at the same time; but communicate with each other infrequently

National HPC HECToR– Mechanical properties of proteins in the skin– Large scale parallel simulations (up to 4000 cores)

on more than 1 million atoms.

Why Bother?

• Ultimately use of e-Infrastructure has boosted our productivity– Got research done that otherwise we couldn’t do– Had more papers published in high impact-factor

journals– Won more grants (need to show you have the

resources available and experience of using them)IF > 12 IF > 4IF > 5

Pamela Greenwell, Life SciencesDiscovery of inhibitors for protozoan glycosidases

• Trichomonas vaginalis (TV) is a major co-factor for the acquisition and transmission of HIV, more than 200 million women worldwide affected

• Only 1 drug used in therapy and resistance is a problem

• Glycosidases of TV ,required for the break down the sugar-rich mucin of the urogenital tract, may provide novel drug target

Problems: no recombinant proteins (active), difficult to purify native enzyme, no crystal structure, limited homology to other enzymes (often less than 30%)

Answers

• Use in silico modelling, energy minimisation and docking to investigate ligand binding

• Develop library of structures (more than 2 million) to screen for novel inhibitors

• Design a method of searching the library for chemicals with similar “fingerprints”

• Screen the library in 1 or 2 days

• Use “wet lab” to validate candidates

Requirements: Interfaces, Portals, Grids and Clouds

• Required good collaboration between computer scientists and biologists

• Interfaces developed to enable biologists to use complex programs without knowledge of computer language

• Portals simplify submission and data retrieval

• Grids/ Cloud resources needed to facilitate parallel screening, cutting time required from weeks or months to hours or days

Results• We have already identified a potential inhibitor and tested it

in vitro- and it worked BUT not as well as we had predicted

• Research revealed that the compound was very lipophilic and hence might have reacted with the membrane-associated enzyme

• We have interrogated our library and pulled out a related but less lipophilic compound which binds almost as well in silico

• “Wet-lab” testing will begin when I return from holiday next week- watch this space!!

Phil Fowler, Molecular dynamics