Databases of Neuronal and SubNeuronal Tomographic Structure and Molecular Distribution
A Computer Science Perspective
Maryann Martone Bertram Ludäscher Amarnath Gupta
University of California San Diego
Objectives
Create Databases for
– Cellular and subcellular morphology – Molecular distributions in cellular and subcellular structures – Physiological responses that reveal functional properties of single cells/structures and cellular/subcellular environments
Provide Facilities to
– Perform ad hoc queries – Navigate seamlessly across experiment boundaries – Compute, analyze and compare properties across
• • • • Resolution levels Experimental conditions Cell populations Species populations
Morphological Data
correlating across resolution levels
Light Microscopy
Electron Tomography
compartments: logical zones of invariance
Modeling Morphology
• Quantitative Analysis
– Measurement data is represented in a semistructured manner (XML) – An example: compute the distribution of spine
length, surface area and volume for spiny dendrites from the neostriatum
• The bigger question:
– Can there be a generic conceptual model to express morphological data? – Needs to represent, manipulate, query
• 3D shapes as thick manifolds and non-simple polyhedra • Branching structures • Topological and metric properties
Modeling Molecular Distributions
• A property distribution is
– a histogram of a vector of observations – a histogram of observations arranged over a mesh of surface nodes – a collection of histograms arranged over some anatomic parcellation
• Our current model
– increasingly finer anatomic regions constitute a part_of (conversely has_a) tree – distributions are histograms over the has_a tree
• Next step
– Extend to intra-region distributions over region surfaces and volumes
Integrating over Models
• The idea
– Different models capture correlated but distinct aspects of biological reality – How can we express and evaluate queries that compute data across models
• Our approach
– For each source create a knowledge-base of the anatomy of the observations – Attach the data of each source to the respective knowledge-base – “bridge” the sources by a simple ontological mapping – Compute the query across bridged sources
Demonstrations