Transcript UCSD Neuron-Centered Database

UCSD Neuron-Centered
Database
Amarnath Gupta
Bertram Ludäscher
Maryann Martone
What is Neuron-Centering
(AKA The Holy Grail)
?
• Designing a database system such that it can be used to
represent, store and access
–
–
–
–
–
–
Any property, measurements, …
of Any Nerve Cell or its constituent parts
from Any part of the brain
acquired through Any experiment
at Any spatial resolution
located at Any physical site
in a way that any biologist and biological applications
can use or interface with it
Designing the database
• Three problems
– Modeling the neuronal structure
• To what level of detail?
– Modeling correlated information building on the
neuronal structure
• Structured as complex graphs
– Integrating heterogeneous data (a short detour)
•
•
•
•
Quantitative morphology
Protein localization
Time-series study from physiological experiments
…
• Current Schema (and evolving ..)
Integration through Mediation
User Query
Mediator
Mediator’s query language
XML documents
XML View(s)
XML View(s)
XML View(s)
Wrapper
Web Site
Wrappers also export:
1. Schemas & Metadata
2. Description of
supported queries...
Wrapper
Database
Image
Features
and back
The Knowledge-Base
• Situate every data object in its anatomical
context
– a programmable knowledge-base that integrates
and correlates every observed piece of data
– An illustration
– New data is registered with the knowledge-base
– Insertion of new data reconciles the current
knowledge-base with the new information by:
• Extending the knowledge-base
• Creating new views with complex rules to encode
additional domain knowledge
Query Processing
• Query Types
– Exploratory queries
– Ad-hoc queries
• Our current approach
– Databases and knowledge-bases are integrated through a
mediator built using a deductive database
– Many queries such as protein localization need complex
grouping of data across the nodes of the knowledge-base
– We support some “traversal” queries on graph of data
and knowledge entities
– Painted Neurons as maps: exploring XML/VML-based
interfaces (Ilya Zaslavsky, SDSC, UCSD)
Next Steps
• Modeling
–
–
–
–
–
Maturing the schema
More data types
Richer knowledge-base constructs (e.g. has-part-of)
Connecting with atlases as spatial data objects
Integration with SDSC’s large-scale distributed data
handling system
• Querying
– Capabilities to handle more generic graph queries
– Better integration of pure querying with other
functionality such as statistical computation
– More expressive query interfaces