Peer-to-peer bioinformatics iCapture

1
Peer-to-peer bioinformaticsiCapture Lunch and
Learn November 2004 Stephen Montgomery
2
Searching for Regulatory Variation
3
GOAL

• To facilitate context-driven hypotheses on the
  function of non-coding polymorphisms
• Utilizing diverse annotation
• State-of-the-art regulatory analysis tools
• Characterized regulatory regions and rSNPs

4
Utilizing diverse annotation
p53 alignment
snp density
5
State-of-the-art regulatory analysis tools

• How do we get the best tools into our platform to
  help decode regulatory modules.
• Storing all the results of various predictions
  across a genome can be unwieldy.
• How do we facilitate access to the information
  encoded in the MANY bioinformatics tools?

6
How are bioinformatics tools developed?
Data and Innovation
Requirements
Biologists
Bioinformaticians
Computer Scientists
Innovation
Organization and Interpretation
7
Each community has its own culture
Biologists
Bioinformaticians
Computer Scientists
8
An interaction map of biologists and
bioinformaticians
Bioinformaticians
Biologists
9
Things get more complicated

• Each individual has
• Access to different resources
• Computational / Monetary / Personnel
• Finite time available
• A different social network
• Professional obligations
• Each group
• Organizational boundaries
• Toolkit (suites and scripts)
• Method of providing tools (OS, Internet,
  Interfaces)

10
An improved interaction map of biologists and
bioinformaticians
Bioinformaticians
Biologists
Improved Communication
Access to communities Access to resources Retain
sub-organization
11
A community-based approach to bioinformatics
analysis

• Use the principles of peer-to-peer technology
• Allow biologists to easily discover and run
  bioinformatics tools
• Create a dynamic, reliable network for analysis
• Reduce overlapping integration efforts
• Improve communication within/outside
  organizations
• Address problems relevant to bioinformatics
• Attribution
• Resource distribution
• Specialized data

12
discover and run jobs here or through bioperl
13
(No Transcript)
14
Chinook Architecture
15
(No Transcript)
16
Architecture Outline

• Chinook Configuration
• Chinook Data

17
Chinook Configuration

• Add new services
• Set Server Mode
• Open ports
• Customize Advertisements
• Run

18
Adding services (XML-Based)
19
Adding services (GUI-based)
20
Chinook Server ModesRMI vs. Web services

• Web Services
• Runs over Tomcat and Axis (SOAP engine)
• Harder to set-up, More management features
• Language independent
• RMI
• Uses the RMI registry (included in JDK/JRE)
• Easier to set-up
• Only allows Java to Java communication

21
How the P2P works JXTA Advertisements
22
Customizing your advertisements

• Edit the advertisements/ directory in your
  Chinook installation to include your endpoint
  location.

23
Running Chinook

• Use Ant
• ant p2p-start
• ant server-start
• From an IDE
• From an Installer

24
Chinook Data
Adding new data objects to services Defining Data
objects Providing Database support Client
Interaction
25
Chinook Data / Databases

• Each application defines its own data objects
• ltdata_entry_setgt
• ltnamegtdna_sequencelt/namegt
• ltmaximum_countgt2lt/maximum_countgt
• ltminimum_countgt2lt/minimum_countgt
• ltdata_entry_type_namegtDNA_LOCATIONlt/data_entry_ty
  pe_namegt
• ltdata_entry_type_namegtDNA_FILElt/data_entry_type_n
  amegt
• ltset_output_class_namegtca.bcgsc.chinook.parsing.s
  etoutput.impl.DataEntrySetOutputterImpllt/set_outpu
  t_class_namegt
• lt/data_entry_setgt

26
Data Interpretation

• ltdata_entrygt
• ltdata_entry_type_namegtDNA_LOCATIONlt/data_entry_ty
  pe_namegt
• ltdata_entry_class_namegtca.bcgsc.chinook.common.da
  taentry.objects.impl.SequenceCoordinatelt/data_entr
  y_class_namegt
• ltdata_entry_validation_class_namegtca.bcgsc.chinoo
  k.common.dataentry.validation.impl.SequenceCoordin
  ateValidationlt/data_entry_validation_class_namegt
• ltdata_entry_gui_support_class_namegtca.bcgsc.chino
  ok.common.dataentry.guisupport.impl.SequenceCoordi
  nateGUISupportlt/data_entry_gui_support_class_namegt
  
• ltdata_entry_loader_class_namegtca.bcgsc.chinook.co
  mmon.dataentry.loader.impl.SequenceCoordinateLoade
  rlt/data_entry_loader_class_namegt
• lt/data_entrygt

27
Database support

• Avoiding empty text boxes
• ltdatabasegt
• lttypegtENSEMBLlt/typegt
• ltparsing_classgtca.bcgsc.chinook.server.database.i
  mpl.EnsemblDatabaseTypelt/parsing_classgt
• ltconnectiongt
• ltnamegtdbx440lt/namegt
• lthostgtensembldb.ensembl.orglt/hostgt
• ltusergtanonymouslt/usergt
• ltportgt3306lt/portgt
• lt/connectiongt
• ltminimum_versiongt20lt/minimum_versiongt
• lt/databasegt

28
Database support

• Finding DNA_LOCATION
• public interface DNASequenceDatabase
• public static final String DATA_TYPE
  "DNA_LOCATION"
• /
• Get the DNA sequence for the genomic location
• _at_param species, the species
• _at_param version, the version
• _at_param chromosome, the chr number or contig
• _at_param abs_start, absolute start - genomic
  coordinate
• _at_param abs_end, absolute end - genomic
  coordinate
• _at_param strand, "-1", or "1"
• _at_return the DNA sequence
• _at_throws DatabaseConnectionException, if cannot
  connect to database
• _at_throws SequenceNotFoundException, if cannot
  obtain sequence
• /

29
Client Interaction
30
Use cases of Chinook

• Grid/Cluster computing.
• Internally connect teams / individuals.
• Collaborate with remote individuals.
• Provide an API layer to your algorithms.
• Insert bioinformatics analysis into applications.
• Show off your tools.

31
Algorithms integrated into Chinook
ClustalW Genscan
Conreal Sim4
DIALIGN MSCAN
LAGAN ANN-Spec
Mauve Recursive Gibbs Motif Sampler
ORCA MEME
Shuffle-LAGAN Motifsampler
T-Coffee RSAT oligo analysis
Promoterwise STUBB
Primer3 Teiresias
Eponine wConsensus
ELPH ContigMerger
32
Projects involving Chinook

• OrthoSEQ plans to provides analysis through the
  Chinook/Bioperl Perl API.
• Sockeye uses Chinook to deliver state-of-the-art
  alignment, PCR prediction, and regulatory
  analysis
• Pegasys plans to provide pipeline management to
  subset of services advertised by Chinook.
• Bio-Linux planning to integrate a subset of their
  algorithms.

33
Future Plans
34
Acknowledgements

• GENOME SCIENCES CENTRE
• Steve Jones
• Tony Fu
• Jun Guan
• Keven Lin
• Asim Siddiqui
• Genereg team _at_ GSC
• Mark Mayo
• Bernard Li

• CENTRE FOR MOLECULAR MEDICINE AND THERAPEUTICS
• Wyeth Wasserman
• Jonathan Lim
• UBC BIOINFORMATICS CENTRE
• Francis Ouellette
• Graham McVicker
• Sohrab Shah

Funding MSFHR, Genome Canada VISIT
http//smweb.bcgsc.bc.ca OR http//www.bcgsc.bc.ca
/chinook