Introduction of a Grid Approach in the Biotechnology Industry

1
Introduction of a Grid Approach in the
Biotechnology Industry

• Grid Day
• University of Cyprus, NicosiaMarch 26, 2003

2
BioGrid team Laboratory of software engineering
internet technologies University of Cyprus

• Head of bioGrid team
• Prof. George Papadopoulos george_at_cs.ucy.ac.cy
• Research group
• Aristos Stavrou cs98sa2_at_cs.ucy.ac.cy
• Dr. Dimitrios Vogiatzis dimitrv_at_cs.ucy.ac.cy

3
BioGrid

• 2 year trial IST project
• Sep.2002-Sep.2004
• Project Leader ZooRobotics
• Site www.bio-grid.net

4
Scientific Objectives

• BioGrid is a trial IST project with the following
  objectives
• Development and Integration of grid technologies
  so that
• Researchers obtain an efficient information
  output
• Three tools to be integrated
• PSIMAP, protein interaction discovery
  visualisation
• Space Explorer, gene protein visualisation
• Classification Server, text data mining
• Tools access Information resources
• Databases (protein, gene expression).
• Unstructured data (pubmed abstracts)
• Software tools (TOPS, for protein structural
  comparison)

5
Business Objectives

• Information Grid for large proteomics and
  genomics databases
• Efficient transnational enterprise collaboration
• Faster time to market biotech innovations
• Software license model for bioGrid
• Targeted customers
• Pharmaceutical Aventis, GSK, Novartis, AkzoNobel
• SMEs KeyGene, Inpharmatica, LionBioScience,
  Avantium

6
Data sources and formats I

• Biological objects are complex not standard
  format (XML, ASN.1, proprietary)
• Literature unstructured, text data mining is
  necessary

• Protein Structure
• Protein Data Bank (PDB)
• Structural Classification of Proteins (SCOP,
  CATH)
• Gene expression databases
• Stanford Microarray data (SMA)
• Gene Expression Ominbus
• Biomolecular interaction database (ASN.1 format)
• PubMed (Biology oriented journal abstracts)
• TOPS (protein comparison tool)

7
Data sources and formats II
LITERATURE DATABASE FORMAT XML BASED
(PUBMED) ltPubmedArticlegt ltPMIDgt12649791lt/PMIDgt ltYe
argt2003lt/Yeargt ltJournalgt ltISSNgt0941-3790lt/ISSNgt
lt/Journalgt ltArticleTitlegtNutritional ecology
chances . services to shape procedures lt/Article
Titlegt ltAbstractgt ltAbstractTextgtNutrition
ecology is the science that studies the impacts
of human nutrition on nutrition ecology in
their projects. lt/AbstractTextgt lt/Abstractgt
lt/PubmedArticlegt
PROTEIN DATABASE FORMAT (PDB) Columns denote
attribute e.g. near the end of file, cols 31-54
denote atomic coordinates (3d vector)
GENE EXPRESSION FORMAT (SMD)
8
Tool 1 Space Explorer

• high-dimensional data mapped into a 1, 2 or 3
  dimensional subspace
• interactive web-enabled, virtual reality
  environment
• 3D visualisation complemented by hierarchical
  clustering
• subsequent visualisations as dendrograms, which
  are linked to the scatter plots
• Space Explorer facilitates visual data mining.

9
Tool 2 PSIMAP

• PSIMAP is the first complete protein structural
  domain interaction map
• shows, what kinds of protein domains are found to
  be interacting structurally.
• PSIMAP has specific shapes reflecting the types
  of protein domains their interaction partners

10
Tool 3 Classification server

• custom hierarchy from a sample of documents,
  producing a global, consistent view
• automatic classification of textual items into a
  hierarchy of topics
• Classifications are output in XML for flexible,
  standard data interchange
• APIs for easy integration with existing
  applications and new services

11
Current situation in Biotechnology

• Genetic sequencing databases
• More than 10,000,000 in Genbank
• Protein databasesgt1,000,000 (PIR)
• Tools gt500 on-line
• PubMed 11,000,000 abstracts on-line
• There is not a universal access to data objects
• Not possible to pass automatically
• from gene expression ?protein ?protein families
  ?visualisation

12
BioGrid project overview
13
Requirements for Grid I
scenario genes ? expression ? proteins ?
relevant literature

• Unified view of the data objects (XML) ? unified
  interfaces for users and applications
• Co-operatation of the (gene expression, protein
  interaction connection to literature)
• Passing data objects to the 3-tools

Lower level functionality

• Addition of new resources transparent
• Continuous operation when adding data sources or
  in the event of component failure
• Functionality on a different platforms (unix
  windows)
• Possibility of local caching and net traffic
  regulation (to be determined in the evaluation
  phase)

14
Requirements for Grid II
Requirements lead to the following Grid types

• Information grid accessibility to sources of
  information and tools for analysis
  visualisation
• Knowledge grid data mining machine learning
  for filtering literature abstracts.

15
Schematic design integrating 3 platforms data
sources
PSIMAP
Space Explorer
Classification Server
XML
GRID
Text Mining
16
bioGrid part of health Grids

• Health Grid Projects
• MammoGrid (databases of mammograms)
• GEMSS (Grid Enabled Medical Simulation Services.
  Access to advanced simulation and image
  processing services (www.gemss.de)
• BioMody (discovery and distribution of bio-data
  to web)
• Tambis single user interface to bio-data
• SRS access to unstructured data

• Grid technologies under investigation
• Globus
• Legion by Avaki
• SDSC Storage Resource Broker