In silico docking on grid infrastructures Jean Salzemann LPC of Clermont-Ferrand, France (CNRS/IN2P3) Embrace Workshop, Helsinki, 2006/06/17 Credit : Nicolas Jacq, Vincent Breton

Content WISDOM initiative Challenges of the high throughput virtual docking Development of a grid environments for a large-scale deployment Achieved deployment on EGEE infrastructure Wide In Silico Docking On Malaria Accelerate drug design against H5N1 neuraminidase Perspectives mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

WISDOM initiative WISDOM initiative aims to demonstrate the relevance and the impact of the grid approach to address drug discovery for neglected and emerging diseases. First achieved experiences: Summer 2005: Wide In Silico Docking On Malaria (WISDOM) Spring 2006: Accelerate drug design against H5N1 neuraminidase Partners: Grid infrastructures: EGEE, Auvergrid, TWGrid European projects: Embrace, BioinfoGrid, Share, Simdat Institutes and association: Fraunhofer SCAI, Academia Sinica of Taiwan, ITB, Unimo University, LPC, CMBA, CERN-ARDA, HealthGrid mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

There is a need… to develop new drugs for the diseases of the developing world HIV/AIDS, malaria and Tuberculosis account for 5,6 million deaths Permanent necessity to develop new drugs to fight emerging resistance to drugs (malaria) Unchanged pharmacopeia for decades against trypanosomiasis, leishmaniasis, Chagas disease, ... to be able to develop quickly new drugs against emerging diseases H5N1, SRAS, dengue… are recent examples of emerging diseases Many factors like world-wide exchanges can help propagation of such diseases at a large scale Necessity to quickly adapt to emerging resistances mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Combinatorial libraries Phases of a pharmaceutical development Molecular Docking: Predict how small molecules, such as substrates or drug candidates, bind to a receptor of known 3D structure Target discovery Lead discovery Target Identification Target Validation Lead Identification Lead Optimization Clinical Phases (I-III) The development of a new drug in pharmaceutical research is extremely time consuming and expensive. On the average the complete research process from the early screening experiments to the clinical testing takes 10 to 15 years and costs about 150-300 million dollars. Recent efforts to minimize development time and cost led to automatisation and miniaturization of modern drug testing systems. Through the rapid progress in the fields of combinatorial chemistry and high-throughput screening (HTS) there is an increasing need for interpretation of large amounts of "noisy" data.   Today it is possible to test the binding affinity of several ten thousands potential drug candidates per day. The results of these experiments are analyzed with the help of special computer software (CADD). vHTS Similarity analysis Database filtering Computer Aided Drug Design (CADD) de novo design diversity selection Biophores Alignment Combinatorial libraries ADMET QSAR Duration: 12 – 15 years, Costs: 500 - 800 million US $ mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Grid-enabled virtual screening workflow Grid service customers Chemist/biologist teams Biology teams Data access for expert teams in the world Grid infrastructure Check point Check point Check point Selected hits Hits Target MD services Docking services Annotation services Grid service providers Chimioinformatics teams Bioinformatics teams mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Challenges for high throughput virtual docking Example: data challenge against H5N1 NA Millions of chemical compounds available in laboratories In vitro high Throughput Screening 1$/compound, nearly impossible 300,000 Chemical compounds: ZINC & Chemical combinatorial library Molecular docking (Autodock) ~100 CPU years, 600 GB data Data challenge on EGEE, Auvergrid, TWGrid ~6 weeks on ~2000 computers In vitro screening of 100 hits Hits sorting and refining Target (PDB) : Neuraminidase (8 structures) mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Issues for the grid-enabled high throughput virtual docking Computer-based in-silico screening can help to identify the most promising leads for biological tests systematic and productive reduces the cost of trail-and-error approach In silico docking is well-fitted for grid deployment CPU intensive application Huge amount of output No communication between tasks Issues of a large scale grid deployment The rate of submitted jobs must be carefully monitored The amount of transferred data impacts on grid performance Grid process introduces significant delays Licensed software requires licenses distribution strategy on grid The rate of submitted jobs must be carefully monitored Robust and fault-tolerant environment Testing services and resources for the application The amount of transferred data impacts on grid performance Install application on grid Subsets of the database instead of large unique compound files Grid process introduces significant delays Submission is limited by the number of Resources Brokers All resources must be filled but saturation must be avoided Licensed software requires licenses distribution strategy on grid mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Grid tools of the data challenges WISDOM a workflow of grid job handling: automated job submission, status check and report, error recovery push model job scheduling batch mode job handling http://wisdom.eu-egee.fr DIANE a framework for applications with master-worker model pull mode job scheduling interactive mode job handling with flexible failure recovery feature http://cern.ch/diane mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

WISDOM components Installer Tester User wisdom_install wisdom_test Set of jobs wisdom_execution Workload definition Job submission Job monitoring Job bookkeeping Fault tracking Fault fixing Job resubmission GRID Grid services (RB, RLS…) Grid resources (CE, SE) Application components (Software, database) Superviser License server Accounting data wisdom_collect wisdom_db wisdom_site mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Simplified grid workflow for WISDOM Results Storage Element Subsets Computing Element WISDOM production system Site1 Statistics Jobs Parameter settings Target structures Resource Broker Computing Element User interface Site2 Subsets Compounds database Storage Element Software Results FlexX license server : 3000 floating licenses given by BioSolveIT to SCAI Maximum number of used licenses was 1008 mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Grid resources of the data challenges EGEE-II AuverGrid TWGrid a world-wide infrastructure providing freely over than 5,000 CPUs and 21 TB for biomedical applications mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

First biomedical data challenge: World-wide In Silico Docking On Malaria (WISDOM) Significant biological parameters 2 different docking applications (Autodock and FlexX) About 1 million virtual compounds selected Target proteins from the parasite responsible for malaria Significant numbers Total of about 46 million ligands docked in 6 weeks 1TB of data produced Up 1700 computers in 15 countries used simultaneously About 80 CPU years Average crunching factor ~600 Number of docked compounds vs time Number of running and waiting jobs vs time mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Second biomedical data challenge: Accelerate drug design against H5N1 neuraminidase Significant biological parameters 1 docking application (Autodock) About 300,000 virtual compounds selected Target proteins with predicted mutations involved in the virus multiplication Significant numbers Total of about 2,5 million ligands docked in 6 weeks 600 GB of data produced Up 2000 computers in 17 countries used simultaneously corresponding to about 105 CPU years Average crunching factor ~900 France,25% UKI,23% SouthWestern Europe,16% SouthEastern Europe,14% Italy,6% Northern Europe,4% Asia Pacific,5% Germany-Switzerland,1% Central Europe,2% Russia,4% Rate of jobs by EGEE federation mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Selecting the promising compounds The in-silico screening provides not only the docking poses of a compound against the target but also the docking energy By ranking the information, chemist can select the promising compounds to go on the structure-based drug design for potential drugs mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01

Perspectives Second large scale docking on EGEE in fall 2006 Several new foreseen targets on malaria, dengue and other neglected diseases. Resources needed: ~80 CPU years per target Supported by EGEE-II and EELA european projects, Swiss BioGrid initiative Collaboration is open for new targets,software & infrastructures Reranking of WISDOM hits by Molecular Dynamics simulations Supported by BioinfoGrid & EGEE-II european projects Interest for ressources on supercomputers (contact with DEISA) Best hits further processed through in vitro testing and structure activity relationships mardi 2 avril 2019 In silico docking on grid infrastructures, NGN Helsinki, 2006/06/01