Grant: 01IG09006 The MoSGrid Portal – A workflow-enabled Grid Portal for Molecular Simulations...
-
Upload
jennifer-galloway -
Category
Documents
-
view
220 -
download
1
Transcript of Grant: 01IG09006 The MoSGrid Portal – A workflow-enabled Grid Portal for Molecular Simulations...
grant: 01IG09006
The MoSGrid Portal – A workflow-enabled Grid Portal for
Molecular Simulations
Sandra GesingCenter for Bioinformatics, University of Tübingen
www.mosgrid.de 2
Outline
• Motivation• MoSGrid (Molecular Simulation Grid)• The MoSGrid portal• Domain specific workflows• MSML (Molecular Simulation Markup Language)• Future work
MoSGrid Portal
www.mosgrid.de 3
Motivation• Numerous applications for molecular simulations and
docking, e.g. • Materials science• Structural biology • Drug design• Sophisticated tools and algorithms support scientists• High-performance computing facilities are available
MoSGrid Portal
www.mosgrid.de 4
Motivation
Drawbacks of using molecular simulations and docking• Usability of tools is limited• Complexity of methods• Lack of graphical user interfaces• Complexity of infrastructures• Many end users lack computer science background
⇒ Need for self-explanatory and intuitive user interfaces ⇒ A portal for molecular simulations and docking
MoSGrid Portal
www.mosgrid.de 5
Portals
• Single point of entry• Possibility to customize views and tools• Store user preferences• No installation of software on the user’s side • No firewall issues
MoSGrid Portal
www.mosgrid.de 6
Unifying Diversity
12181 acatttctac caacagtgga tgaggttgtt ggtctatgtt ctcaccaaat ttggtgttgt 12241 cagtctttta aattttaacc tttagagaag agtcatacag tcaatagcct tttttagctt 12301 gaccatccta atagatacac agtggtgtct cactgtgatt ttaatttgca ttttcctgct 12361 gactaattat gttgagcttg ttaccattta gacaacttca ttagagaagt gtctaatatt 12421 taggtgactt gcctgttttt ttttaattgg gatcttaatt tttttaaatt attgatttgt 12481 aggagctatt tatatattct ggatacaagt tctttatcag atacacagtt tgtgactatt 12541 ttcttataag tctgtggttt ttatattaat gtttttattg atgactgttt tttacaattg 12601 tggttaagta tacatgacat aaaacggatt atcttaacca ttttaaaatg taaaattcga 12661 tggcattaag tacatccaca atattgtgca actatcacca ctatcatact ccaaaagggc 12721 atccaatacc cattaagctg tcactcccca atctcccatt ttcccacccc tgacaatcaa 12781
taacccattt tctgtctcta tggatttgcc tgttctggat attcatatta atagaatcaa
Slide copied from: Stuart Owen „Workflows with Taverna“
MoSGrid Portal
www.mosgrid.de 7
MoSGrid
Molecular Simulation Grid (D-Grid project)Goal
Providing users with Grid access to molecular simulation tools and docking tools via a workflow-enabled portal
• Implementation of high-performance computing• Workflows• Annotations of results• Data mining• Use of the D-Grid-infrastructure
MoSGrid Portal
www.mosgrid.de 8
MoSGrid Partners
• Universität zu Köln• Eberhard-Karls-Universität Tübingen• Universität Paderborn• Konrad-Zuse-Zentrum für Informationstechnik
Berlin• Technische Universität Dresden• Technische Universität Dortmund• Bayer Technology Services GmbH, Leverkusen• Origines GmbH, Martinsried• GETLIG&TAR, Falkensee• BioSolveIT, Sankt Augustin• COSMOlogic GmbH&Co. KG, Leverkusen
MoSGrid Portal
www.mosgrid.de 9
MoSGrid in a Nutshell
XtreemFS
CloudFile
System
PortalWS-PGRADE
Grid resourcesUNICORE 6
Result
RecipeStructure Result
High-level middleware service level
gUSE
Workflow
MoSGrid Portal
www.mosgrid.de 10
Credential Management
• User management based on Liferay features- Community management- Organization management• X.509 user certificates • SAML (Security Assertion Markup Language)- Minimize credential data transfers - Set of maximum hops for trust delegation- Usable for single sign-on infrastructures (e.g.,
Shibboleth)
MoSGrid Portal
www.mosgrid.de 11
Credential Management
MoSGrid Portal
www.mosgrid.de 12
WS-PGRADE
MoSGrid Portal
www.mosgrid.de 13
WS-PGRADE
MoSGrid Portal
www.mosgrid.de 14
WS-PGRADE
MoSGrid Portal
www.mosgrid.de 15
gUSE Architecture
User interfaceWS-PGRADE
Grid resources middleware layer
UNICORE 6
Applicationrepository
Information system
LoggingSubmitters
Workflow storage
Workflowengine
High-level middleware service layer
gUSE
grid User Support Environment
MoSGrid Portal
www.mosgrid.de 16
gUSE Submitter
Interface GridService• actionJobSubmit• actionJobAbort• actionJobOutput• actionJobStatus• actionJobResource
JOBn
Workflowengine
JOB1
JOB2JOB3 JOB4
SubmitterGridService
MoSGrid Portal
www.mosgrid.de 17
gUSE Submitter for UNICORE
JOBn
Workflowengine
JOB1
JOB2JOB3 JOB4
UNICORE submitter(UCC lib)
UNICORE Atomic
Services
Uspace
gUSE UNICORE 6 Resources
4 - Upload data
1 - Security2 - Registry3 - Submit job5 - Start job
actionJobSubmit
MoSGrid Portal
www.mosgrid.de 18
ASM (Application Specific Module)• Library for managing WS-PGRADE workflows • Listing of users and workflows in the local repository• Import of Workflows in the user space• Upload/download of input and output files• Setting the parameters of a job in a workflow• Submission of workflows• Monitoring of workflows• Deletion of workflows
• Usable in portlets und Java tools ⇒ Implicit use of gUSE submitter
MoSGrid Portal
www.mosgrid.de 19
• XtreemFS is an object-based grid and cloud filesystem• Ability to minimize data transfer• Low latency,
local availability through replication
• Grid Security Infrastructure (GSI) support
Distributed Data Management
MoSGrid Portal
www.mosgrid.de 20
• XtreemFS integration • Portlet• UNICORE• GSI support
• Data flow • WS-PGRADE• XtreemFS • Frontend nodes • Compute nodes
• UNICORE mediates data transfers
XtreemFS
UNICORETSI
Distributed Data Management
MoSGrid Portal
www.mosgrid.de 21
Domain Molecular Dynamics
• Study and simulation of molecular motion• Provide a molecular dynamics service on multiple
levels• Direct upload of job descriptions• Workflows and standard recipes for repeating
tasks• Analysis of relevant properties
MoSGrid Portal
www.mosgrid.de 22
Equilibration of Proteins
• Proteins from databases (e.g., the Protein Data Bank, PDB) do not necessarily represent a near-native conformation/configuration
• For all kind of production runs a minimization and an equilibration is an indispensable prerequisite
• Eases the work of experienced users• Lowers the hurdle for novice users
MoSGrid Portal
www.mosgrid.de 23MoSGrid Portal
UseCase: Gromacs_EQ
structure(pdb/gro)
topology(top/itp)
EM.mdp(mdp)
pdb2gmxstructure
(pdb)
editconf
box(pdb)
genbox
Solvated(pdb)
grompp
adj. Top.(top/itp)
topol.tprmdout.m
dp
www.mosgrid.de 24MoSGrid Portal
mdrun
ener.edr traj.trr traj.xtc
md.logstate.cp
tSYSTEM_EM.pdb
grompp
mdrun
topol.tprmdout.m
dp
ener.edr traj.trr traj.xtc
md.logstate.cptSYSTEM_E
Q.pdb
FULL.mdp
(mdp)
g_energyxmgrace
Analysis.jpg
g_energyxmgrace
Analysis.jpg
www.mosgrid.de 25
MD Portlet
MoSGrid Portal
www.mosgrid.de 26
Domain Quantum Chemistry
• Study and simulation of molecular electronic behavior relative to their chemical reactivity
• Survey - MoSGrid Community• First implementation for Gaussian• Then support for• Turbomole• GAMESS-US• Further relevant QC applications
MoSGrid Portal
www.mosgrid.de 27
Domain Quantum Chemistry
Gaussian Jobs• Single input file• Defines molecular geometry and task
• Result• Not structured output• Platform dependent checkpoint file
• Integrated multi-step job option• Not usable for generalized workflows
MoSGrid Portal
www.mosgrid.de 28
Domain Quantum Chemistry
First prototype• Workflow controlled by portlet• Three phases • Pre-processing• Job execution• Post-processing
MoSGrid Portal
www.mosgrid.de 29
Domain Quantum Chemistry
Assisted job creation• Guiding GUI• Most common options
available
Pre-created job description• Upload of Gaussian job
description file
Monitoring of jobsPost-processing and presentation of results
Workflows
MoSGrid Portal
www.mosgrid.de 30
Domain Quantum Chemistry
Preprocessing• Portlet (GUI) supports common options• Automatic generation of job description• Submission
of job
MoSGrid Portal
www.mosgrid.de 31
Domain Quantum Chemistry
Post-processing• Parsing of result file• Python scripts executed by portlet• Relevant information about molecular properties
• Data in CSV-Format saved and accessible
MoSGrid Portal
www.mosgrid.de 32
Domain Docking
• CADDSuite (Computer-aided Drug Design)
MoSGrid Portal
www.mosgrid.de 33
• Galaxy available for local ressources in Tübingen
Domain Docking
MoSGrid Portal
www.mosgrid.de 34
MolDB
• Stores molecules in binary format, which allows for fast export• Automatically creates and stores can. smiles, fingerprints, and functional groups counts for imported molecules• Automatically saves and restores docking-/rescoring-results• DB can be filtered to all stored molecule properties before exporting molecules• Current speed for import/export: ~100 compounds/sec.
MoSGrid Portal
www.mosgrid.de 35
MSML
Molecular Simulation Markup Language• Based on CML (Chemical Markup Language)• Common interpretation by humans and
computers• Follows the minimum information principle• Description:
http://xml-cml.org/convention/dictionary• XSL transformation• Used for validation purposes
validator.xml-cml.orgMoSGrid Portal
www.mosgrid.de 36
Future Work
• WS-PGRADE• Integration of the UNICORE IDB to offer drop-down
boxes of available tools• MD- and QC-Portlet• Adoption to gUSE workflow engine via the ASM
libraries• CADDSuite• Export of workflows from Galaxy to WS-PGRADE
• MSML• Further development
MoSGrid Portal
www.mosgrid.de 37
Involved Projects
SHIWA (SHaring Interoperable Workflows for Large Scale Scientific Simulations on Available DCIs)
• EU project• Duration: 01.07.2010 – 30.06.2012• Tübingen participates via Galaxy workflow export
CompChem Virtual Organization• EGEE project• Available ressources
MoSGrid Portal
www.mosgrid.de 38
Future Projects
SCI-BUS (SCIentific gateway Based User Support)• EU project• Duration: 01.10.2011 – 30.09.2014• Pan-European ressources• Tübingen participates with the extension of the
MoSGrid portal with an interactive molecule editor and a semantic search
MoSGrid Portal
www.mosgrid.de 39
Acknowledgements• Oliver Kohlbacher• Ákos Balaskó• Georg Birkenheuer • Sebastian Breuers• Richard Grunzke• Sonja Herres-Pawlis • Valentina Huber• Miklos Kozlovszky • Jens Krüger• István Márton • Patrick Schäfer• Bernd Schuller• Johannes Schuster• Anna Szikszay Fabri• Klaus-Dieter Warzecha• Martin Wewior
MoSGrid Portal
www.mosgrid.de 40MoSGrid Portal