MTA SzTAKI - 27.08.2010 (pdf) - Universität Tübingen
MTA SzTAKI - 27.08.2010 (pdf) - Universität Tübingen
MTA SzTAKI - 27.08.2010 (pdf) - Universität Tübingen
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
A gUSE submitter for<br />
MoSGrid<br />
(Molecular Simulation Grid)<br />
Sandra Gesing<br />
sandra.gesing@uni-tuebingen.de<br />
Eberhard-Karls-<strong>Universität</strong> <strong>Tübingen</strong><br />
<strong>27.08.2010</strong>
Outline<br />
• Molecular simulations<br />
• Users<br />
• MoSGrid<br />
• MoSGrid architecture<br />
• gUSE submitter<br />
• Future work<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong> 1
Molecular simulations<br />
• Computational techniques<br />
• Predicting many useful functional properties of<br />
chemicals and materials<br />
• Thermodynamic properties<br />
• Thermochemical properties<br />
• Spectroscopic properties<br />
• Mechanical properties<br />
• Transport properties<br />
• Morphological information<br />
• Diverse fields of applications, e.g. drug design<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
2
Molecular simulations<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
http://www.nmr.chem.uu.nl/haddock2.0/movies/cg1.html<br />
3
Users<br />
• Sophisticated tools and algorithms available<br />
• Scientists with different computational<br />
background<br />
• Need of self-explanatory and intuitive user<br />
interfaces<br />
⇒ Portals<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
4
Unifying diversity<br />
12181 acatttctac caacagtgga tgaggttgtt ggtctatgtt ctcaccaaat ttggtgttgt<br />
12241 cagtctttta aattttaacc tttagagaag agtcatacag tcaatagcct tttttagctt<br />
12301 gaccatccta atagatacac agtggtgtct cactgtgatt ttaatttgca ttttcctgct<br />
12361 gactaattat gttgagcttg ttaccattta gacaacttca ttagagaagt gtctaatatt<br />
12421 taggtgactt gcctgttttt ttttaattgg gatcttaatt tttttaaatt attgatttgt<br />
12481 aggagctatt tatatattct ggatacaagt tctttatcag atacacagtt tgtgactatt<br />
12541 ttcttataag tctgtggttt ttatattaat gtttttattg atgactgttt tttacaattg<br />
12601 tggttaagta tacatgacat aaaacggatt atcttaacca ttttaaaatg taaaattcga<br />
12661 tggcattaag tacatccaca atattgtgca actatcacca ctatcatact ccaaaagggc<br />
12721 atccaatacc cattaagctg tcactcccca atctcccatt ttcccacccc tgacaatcaa<br />
12781 taacccattt tctgtctcta tggatttgcc tgttctggat attcatatta atagaatcaa<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
Slide copied from: Stuart Owen „Workflows with Taverna“<br />
5
MoSGrid<br />
Molecular Simulation Grid (D-Grid project)<br />
Goal<br />
• Providing users with Grid services for molecular<br />
simulation tools and docking via a portal<br />
• Implementation of high-performance computing<br />
• Workflows<br />
• Annotations of results<br />
• Data mining<br />
• Use of the D-Grid-infrastructure<br />
6
MoSGrid - Partners<br />
• <strong>Universität</strong> zu Köln<br />
• Eberhard-Karls-<strong>Universität</strong> <strong>Tübingen</strong><br />
• <strong>Universität</strong> Paderborn<br />
• Konrad-Zuse-Zentrum für Informationstechnik<br />
Berlin<br />
• Technische <strong>Universität</strong> Dresden<br />
• Bayer Technology Services GmbH, Leverkusen<br />
• Origines GmbH, Martinsried<br />
• GETLIG&TAR, Falkensee<br />
• BioSolveIT, Sankt Augustin<br />
• COSMOlogic GmbH&Co. KG, Leverkusen<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
7
MoSGrid - Community<br />
1. Technische <strong>Universität</strong> Chemnitz, Fakultät für Naturwissenschaften, Institut für Chemie<br />
2. International Center for Information Technology, Rheinische Friedrich-Wilhelms-<strong>Universität</strong> Bonn<br />
3. Center of Bioinformatics Saar, <strong>Universität</strong> des Saarlandes<br />
4. Institut für Physikalische und Theoretische Chemie,Technische <strong>Universität</strong> Braunschweig<br />
5. Bremen Center for Computational Materials Science, <strong>Universität</strong> Bremen<br />
6. <strong>Universität</strong> Münster, Organisch-Chemisches Institut<br />
7. Max-Planck-Institut für biophysikalische Chemie,Göttingen<br />
8. Otto-Diels-Institut für Organische Chemie, Christian-Albrechts-<strong>Universität</strong> zu Kiel<br />
9. Johann Wolfgang Goethe-<strong>Universität</strong> Frankfurt am Main,Institut für Anorganische und Analytische Chemie<br />
10. Department Chemie, Fakultät für Naturwissenschaften,<strong>Universität</strong> Paderborn<br />
11. Technische <strong>Universität</strong> Braunschweig, Institut für Organische Chemie<br />
12. Institut für Pharmazeutische Chemie, <strong>Universität</strong> Marburg<br />
13. <strong>Universität</strong> Karlsruhe (TH), Institut für Physikalische Chemie<br />
14. Albert-Ludwigs-<strong>Universität</strong> Freiburg, Institut für Anorganische und Analytische Chemie<br />
15. Max-Planck-Institut für Informatik, Saarbrücken<br />
16. Lehrstuhl für Theoretische Chemie, Ruhr-<strong>Universität</strong> Bochum<br />
17. Department Physik, Fakultät für Naturwissenschaften, <strong>Universität</strong> Paderborn<br />
18. Lehrstuhl für Theoretische Chemie, <strong>Universität</strong> Bonn<br />
19. Institut für Physikalische und Theoretische Chemie, <strong>Universität</strong> <strong>Tübingen</strong><br />
20. Zentrum für Bioinformatik, <strong>Universität</strong> Hamburg<br />
21. Lehrstuhl für Biophysik, Ruhr-<strong>Universität</strong> Bochum<br />
22. Department Physik, Fakultät für Naturwissenschaften, <strong>Universität</strong> Paderborn<br />
23. Johann Wolfgang Goethe <strong>Universität</strong>, Fachbereich Chemie<br />
24. <strong>Universität</strong> Rostock, Institut für Chemie, Abteilung für Anorganische Chemie<br />
25. Humboldt <strong>Universität</strong> zu Berlin, Institut für Chemie<br />
26. Institut für Pharmazie, Martin-Luther-<strong>Universität</strong> Halle-Wittenberg<br />
27. Technische <strong>Universität</strong> Braunschweig, Institut für Anorganische und Analytische Chemie<br />
28. Zentrum für Bioinformatik, <strong>Universität</strong> Hamburg<br />
29. Department Chemie, <strong>Universität</strong> Hamburg<br />
30. Fachbereich Chemie - Theoretische Chemie, Technische <strong>Universität</strong> Kaiserslautern<br />
31. Fachbereich Angewandte Naturwissenschaften, Fachhochschule Gelsenkirchen<br />
32. Ludwig-Maximilians-<strong>Universität</strong> München, Department Chemie und Biochemie<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
8
MoSGrid<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
9
MoSGrid Components<br />
• survey about molecular simulation tools<br />
⇒ Gaussian, Gromacs, FlexX (to start with)<br />
• evaluation about portals<br />
⇒ Liferay, WS-PGRADE<br />
• evaluation about workflow engines<br />
⇒ UNICORE, WS-PGRADE<br />
• evaluation about Grid middlewares<br />
⇒ UNICORE<br />
(D-Grid infrastructure GT 4, gLite, UNICORE)<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
10
MoSGrid Architecture<br />
Grid middleware<br />
Workflow engine<br />
(UNICORE 6)<br />
Batch system<br />
WS-PGRADE portal<br />
(integrated workflow editor)<br />
Workflow engine Repository<br />
Services<br />
Hardware<br />
(local, internet or<br />
integrated in a Grid or Cloud infrastructure)<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
Repository<br />
(XtreemFS)<br />
11
gUSE submitter<br />
Installation of WS-PGRADE portal in <strong>Tübingen</strong><br />
• added own VO bwGRiD<br />
• submission to GT 4<br />
⇒ extension of existing documentation<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
12
gUSE submitter<br />
Available information at the beginning<br />
• József’s presentation about experiences with<br />
implementing a gUSE submitter<br />
• source code of the BES interface for P-GRADE in<br />
G3Bridge<br />
• source code of a BES submitter for gUSE (not<br />
working)<br />
• WS-PGRADE installation (without source code)<br />
⇒ Start to implement BES submitter<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
13
gUSE submitter<br />
In order to test your submitter, you need<br />
1. jdk 1.6 [on the testing machine]<br />
2. tomcat [on the testing machine]<br />
3. gUSE jar files for submitter [on the testing machine]<br />
4. your own developed submitter jar file [on the testing machine]<br />
5. appropriate configuration of your submitter [on the testing machine]<br />
6. accessing an existing and running gUSE/WS-PGRADE<br />
7. configuring gUSE/WS-PGRADE to use your submitter<br />
For getting 1: download it from java.sun.com<br />
For getting 2,3,5: gUSE developers can provide a prepared “submitter<br />
package”, only a few parameters must be configured by yourself<br />
For getting 6: get an account from the portal operators<br />
For getting 7: ask the portal operators to add your submitter to gUSE<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
Slide copied from József‘s presentation<br />
14
gUSE submitter<br />
Creation of the environment under<br />
pgportal/tomcat/webapps/submitter/submitter_BES<br />
• copy submitter_local<br />
• edit of<br />
− server-config.wsdd<br />
− web.xml<br />
− service.properties<br />
− service.xml<br />
− init.jsp<br />
• GridService interface<br />
• JobConfig class<br />
• implement Grid_BES<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
15
GridService interface<br />
package hu.sztaki.lpds.submitter.grids.inf;<br />
public interface GridService<br />
{<br />
public void actionJobSubmit(String pPath,JobConfig pJC,<br />
SubmitterPortalClient pProxy);<br />
public void actionJobAbort();<br />
public void actionJobOutput();<br />
public int actionJobStatus();<br />
public String actionJobResource();<br />
}<br />
public void actionSetJobStatus(int pValue);<br />
public void actionSetJobResource(String pValue);<br />
public void setRunnerId(String pValue);<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
Slide copied from József‘s presentation<br />
16
JobConfig class<br />
• JobConfig stores 3 hash tables<br />
– “Job”, “Input” and “Output” properties<br />
• Job properties table stores property-value pairs, where<br />
– “property” is one of the job properties<br />
– “value” is a string associated to the property<br />
– “property” “value”<br />
• e.g.: “params” ”-i input1.inp -o out1”<br />
• e.g.: “resource” “ares02.cyf-kr.edu.pl”<br />
• Input and Output tables store key-property-value triples, where<br />
– “key” is the name of a job port<br />
– “property” is the name the property<br />
– “value” is a string associated to the property<br />
– (“key”, ”property”) (“value”)<br />
• e.g.: (“INPORT1”,”intname”) (”input1.inp”)<br />
• e.g.: (“OUTPORT1”,”intname”) (”out1”)<br />
• The list of properties can be found at:<br />
http://wiki.sztaki.hu/lpds/index.php/T%C3%A1tolt_adatok<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
Slide copied from József‘s presentation<br />
17
gUSE submitter<br />
Use of source code of the provided BES submitter<br />
• source code of GridService and JobConfig<br />
included<br />
• first compilations of Grid_BES with the result of<br />
null pointer exceptions<br />
⇒ GridService and JobConfig not up-to-date any<br />
more<br />
⇒ import both classes from the WS-PGRADE<br />
installation as jar-files<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
18
gUSE submitter<br />
Use of source code of the provided BES interface<br />
for P-GRADE<br />
• in G3Bridge much more information included<br />
than the certificate, e.g. ftp user<br />
• using similar methods for BES submitter but<br />
different protocol (BFT)<br />
⇒ need to know:<br />
The list of properties can be found at:<br />
http://wiki.sztaki.hu/lpds/index.php/T%C3%A1tolt_adatok<br />
(Hungarian comments but with examples)<br />
Java doc (Hungarian)<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
19
gUSE submitter<br />
Some drawbacks occured…<br />
• my portal server was moved and not reachable<br />
for three days<br />
• needed four days to get access to a working<br />
UNICORE 6 installation<br />
• bug in the BES-UNICORE-interface for other<br />
protocols than ftp<br />
⇒ started to implement submitter_unicore<br />
• JobConfig.class in WS-PGRADE installation not<br />
up-to-date<br />
⇒ needed source code of JobConfig.java<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
20
gUSE submitter_unicore<br />
Current version<br />
• setup security (at the moment my certificate)<br />
• create UNICORE registry client<br />
• find an appropriate target system<br />
• create the job and submit it to the target system<br />
• upload executable and input files into the job<br />
directory (Uspace)<br />
• start the job<br />
• download output files (Uspace)<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
21
gUSE submitter_unicore<br />
Current status<br />
• jobs can be started<br />
• UNICORE returns status SUCCESSFUL<br />
• output is empty<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
22
gUSE submitter_unicore<br />
Next steps<br />
• fix the bug with the output<br />
• add SAML security<br />
• add generator availability<br />
• add possibility to submit UNICORE workflows<br />
Last but not least<br />
• English Java doc<br />
• English manual<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
23
Future work<br />
• user-friendly parameter-sweeps<br />
• choosing of executables without uploading<br />
• reporting of intermediate results<br />
• input / output masks for executables<br />
• data management (upload, local, global)<br />
• repository of molecules and recipes<br />
• restrictions on distributing tasks (dependent on<br />
location of the data)<br />
• conversion of UNICORE workflows to WS-<br />
PGRADE workflows<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
24
MoSGrid timeline<br />
13.-14.09.2010 MoSGrid Meeting<br />
(presentation of submitter)<br />
07.-09.11.2010 6. German Conference on Chemoinformatics<br />
http://www.gdch.de/vas/tagungen/tg/5412__e.htm<br />
(MoSGrid will be introduced with link to a<br />
demonstration of the portal; later the link will redirect to<br />
the „real“ portal)<br />
01.12.2010 Portal-Prototype with workflow infrastructure for user<br />
community<br />
01.05.2011 Portal-Prototype with access to molecule data<br />
repository (beta-tester)<br />
01.12.2011 Portal-Prototype with data repository and workflow<br />
infrastructure for user community<br />
01.06.2012 Portal will be released<br />
30.08.2012 End of project MoSGrid<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
25
Thank you for your attention.<br />
Sandra Gesing - A gUSE submitter for MoSGrid – <strong>27.08.2010</strong><br />
26