Using the WS-PGRADE Portal in the ProSim Project Protein Molecule Simulation on the Grid Tamas Kiss, Gabor Testyanszky, Noam Weingarten, Zsolt Lichtenberger 1 st P-GRADE Portal User Community Workshop June 10-11, Zurich, Switzerland

The biological interest The motivation: Understanding how sugars interact with their protein partners may lead to development of new treatment methods for many diseases. The obstacle: Investigation of the binding of proteins to sugars in “wet laboratory” (in vitro) experiments is expensive and time consuming Expensive substrates Sophisticated machinery The solution: Use “in silico” tools (computer simulation) to select best binding candidates In vitro work only on selected candidates

The biological interest Binding pocket Sugar (ligand) Protein (receptor)

The biological interest Advantages of in silico methods: Better focusing wet laboratory resources: Better planning of experiments by selecting best molecules to investigate in vitro Reduced time and cost Increased number of molecules screened Problems of in silico experiments: Time consuming Weeks or months on a single computer Simulation tools are too complex for an average bio-scientist Unix command line interfaces Bio-molecular simulation tools are not widely tested and validated Are the results really useful and accurate?

What can we gain via the simulation? 1. 1.Validation and refinement of in-silico modelling tools 2. 2.Filter potential scenarios for wet lab experiments

The researcher’s interest What does the researcher want? Run the simulations faster use Grid resources – National Grid Service (NGS) and EGEE Run the simulations using seamless access to compute resources web based interface combining many simulation, analysis and visualisation tool workflows running multiple docking experiments and molecular dynamics analysis to investigate different scenarios parameter study

ProSim – Protein Molecule Simulation on the Grid Funded by the JISC- ENGAGE program Engaging Research with e-Infrastructure promote the greater engagement of academic researchers in the UK with the UK's e-Infrastructure Prosim objectives: – –define user requirements and user scenarios of protein molecule simulation – –Identify, test and select software packages for protein molecule simulation – –automate the protein molecule simulation creating workflows and parameter study support. – –develop application specific graphical user interfaces – –run protein molecule simulation on the UK National Grid Service and make it available for the bioscience research community.

The User Scenario PDB file 1 (Receptor) PDB file 2 (Ligand) Energy Minimization (Gromacs) Validate (Molprobity) Check (Molprobity) Perform docking (AutoDock) Molecular Dynamics (Gromacs) Phase 1 Phase 2 Phase 3 Phase 4

The User Scenario in detail Public repository Local database User provided Preparation and standardisation Solvation and charge neutralization Energy minimization Validation phase 1 – selection and preparation of receptor Solvation Energy minimization Built using SMILES Public repository Local database User provided phase 2 – selection and preparation of ligand

The User Scenario Prepare docking: docking parameters and grid-space - AutoGrid Docking and selection of best results according to total energy AutoDock 10 AutoDock executions, 100 genetic algorithm runs each phase 3 – docking ligand to receptor Solvation of the ligand- receptor structure Energy minimisation – GROMACS Molecular dynamics GROMACS MPI version Molecule trajectory data analysis phase 4 – refining the ligand- receptor molecule (performed on 10 best results of the AutoDock simulation)

The Workflow in WS-PGRADE a combination of GEMLCA and standard g-USE jobs Executed on 5 different sites of the UK NGS Parameter sweeps in phases 3 (via a script) and 4 (via WS- PGRADE parameter sweep)

Running simulations Set input parameters Upload input files Select executor sites Follow execution progress Typical execution time: 24 hours

User views Biologist end-user Minimal computer and g-USE skills Only interested in running her own reserach Import, parameterize, execute and visualise workflows only Expert user g-USE and computer literate biologist Modify workflows Design new experiments Communicate end-user request towards IT team

The ProSim visualiser Visualisation in a newly developed portlet Allows visualisation of receptor, ligand and docked molecules at any phase during and after simulation (if the necessary files have already been generated) Allows to visualise and compare two molecules at a time. Energy, pressure, temperature and other important statistics statistics are also displayed. Using the KiNG ((Kinemage, Next Generation) visualisation tool

The ProSim visualiser

The ProSim visualiser

Lessons learned Communication between scientists and Grid experts is extremely difficult More than 50% of total time spent for the project is for communication and describing/understanding user requests/requirements Novice Grid users require totally transparent access to Grid resources User is interested in her science and not in MPI, Globus or WMS.

Future plans Make workflow more flexible to accommodate numerous different user scenarios Investigate further scenarios such as virtual screening of many ligands to one selected receptor Led to follow-up co-operation with Imperial College London: High-throughput Molecular Dynamics Modelling of Molecular Machines - proposal under review

Thank you for your attention! Any questions? Contact and more information: Paper: Tamas Kiss, Pamela Greenwell, Hans Heindl, Gabor Terstyanszky and Noam Weingarten, Parameter Sweep Workflows for Modelling Carbohydrate Recognition, Submitted to the Journal of Grid Computing, Currently under review.