Building a Neuroanatomy

1
Building a Neuroanatomy
Data Collection
Database

• Amarnath Gupta
• Maryann Martone Bertram Ludäscher Mark Ellisman

2
Morphological Data
filled cell
treated cell
cell model

• Correlated image sets
• Quantitative Morphology
• average arborization pattern
• variation across brain regions, species mutants,
  developmental stages
• is this cell different from the others?
• Spatiotemporal properties (signal propagation)

3
Correlated Multiresolution Data

• Both 2D stacks and selective volumetric data
• Measurements made across volumetric data
• Correlation allows search for corresponding
  data
• Aggregate queries such as distribution of
  measured variables

4
Electron Tomography

• Image/Volume data needs to be explicitly modeled
• Cell/organelle grouping and subgrouping to be
  represented
• Topological arrangement of data
• User-defined methods for comparison of groups and
  arrangements

5
Protein Localization

• Localization at multiple resolutions
• Intensity distribution is a measure of
  localization
• Must be connected to proteins, their isoforms and
  mutants from other data sources like PDB,
  Swiss-Prot and CaBP

cell-level
tissue-level
6
Using the NPACI Data Infrastructure

• The images/reconstructed volumes become SRB
  objects to be stored in HPSS
• The application data model
• MCAT with T-language?
• MCAT with Oracle XML support?
• Exelon with XQL?
• MIX with data stored in Files/Orcale?

7
What are we missing?
Data Modeling
Production/Analysis Tools

• Object-orientation
• inheritance, behavior
• complex group-forming properties
• Recursive computation
• structural recursion
• transitive closure
• graph-operations
• Feature-based operations
• histogram comparison
• Inference capabilities (some)

• Standardization in naming and object reference
  for all tools producing data
• Modeling tools for image and volumetric data

8
What are we doing about it?

• A small prototype using FLORA
• F-logic engine on XSB prolog
• XSB has an ODBC and and Oracle link to store
  facts
• More scalable and robust version of FLORA coming
  up
• Will also explore other robust software with
  equivalent functionality
• A web site wrapper that extracts information,
  traverses multiple links and fills in forms
• Proposing a standard protocol for image, volume
  and feature information