Development of a 3D Neuron Centered Database

1
Development of a 3D Neuron Centered Database

• Presenters Maryann Martone
• Amarnath Gupta
• University of California San Diego
• Participants
• Bertram Ludaescher
• Mona Wong
• Mark Ellisman

2
Objectives

• Methods for obtaining large scale 3D
  reconstructions of neurons and the distribution
  of their protein constituents
• Tools for viewing and manipulating tomographic
  datasets
• Database of multiresolution datasets, including
  both morphological and protein localization data

www-ncmir.ucsd.edu/NeuronDB
3
Electron Tomography

• Derivation of 3D structure from 2D projections
  taken at different angles

Hippocampal Synapse 0.5 µm thick section
4
Advantages of Tomography

• Thick sections
• Easier to produce
• Facilitates correlated LM and EM analysis
• Avoids alignment errors and missing sections
• Advantageous for tracking complicated membrane
  systems like the ER
• Improved z resolution
• Important for studying structures below 40 nm

5
Improved Efficiency

• Acquisition of tilt series
• newly implemented custom 2K X2K digital camera
• Remote access to high voltage electron microscope
  (telemicroscopy)
• Backprojection
• Distributed processing of parallelized back
  projection algorithms using Globus.
• Visualization and Analysis
• Java based tools for analysis and visualization
• New tools for automatic segmentation

6
3-D Neuron Morphology
Electron Tomography
Light Microscopy
7
Compartmental Models
Neurolucida Microbrightfield
Injected Purkinje neuron
8
3D Protein Distribution on Identified Neurons
Intracellular injection of phalloidin and Alexa
568 Francisco Capani and Eric Bushong
9
XvoxtraceSegmentation of Tomographic Datasets
Plane Perim Area X Y Z Min Max
Mean SD
Purkinje Cell dendritic spine

• Open-GL compatible
• Java 3D in production
• Added function for determining labeling intensity

F-actin in cerebellum
10
Jdend(Formally Xdend)
Quantitative analysis of contour data generated
in Xvoxtrace Java 3D cross platform and web
accessible Mona Wong Steve Lamont
11
3D- Neuron Centered Database

• Goals
• Make multiresolution datasets available to
  researchers via the web
• Store metadata and results of quantitative
  analyses of morphological and protein-labeled
  images
• Query protein distributions in various neuronal
  types and subcellular compartments
• Compare protein distributions across different
  proteins, species, subcellular compartments,
  brain regions etc.
• Output realistic structural data and protein
  distributions to modeling programs like M-Cell

12
Federation of Brain Data

• NPACI National Partnership for Advanced
  Computational Infrastructure
• Knowledge-guided integration of Neuroscience data
• Amarnath Gupta
• Bertram Ludaescher
• Maryann Martone

13
Plans for Upcoming Year

• Populate database
• Export 3D datasets for large scale simulations
• Integration with 3D brain atlases and PDB
• Protein distributions on 3D models

M-Cell Tom Bartol, Salk Institute Joe Stiles,
Cornell University
14
Neuron Painting

• Axiomap
• Takes XML-wrapped files and converts them into
  VML
• Developed for geographical data on the web
• Ilya Zaslavsky
• San Diego Supercomputer Center

15
Summary

• Understanding and insight into scope of problem
  (cell level database)
• Results
• Useful tools for data analysis and management at
  many levels
• New research tools Navigation through levels of
  the nervous system without a monolithic database
• Interdisciplinary team with leading researchers
  from computer science, information science and
  Neuroscience