Physiological Data and Simulation
Participants
- Eliza Chan
- Esther Gardner
- David Goldberg
- Gwen Jacobs
- Hualou Liang
- Tom Morse
- David Mountain
- Giuseppina Tesco
- Jonathan Victor
Problems for Physiological Data & Simulation
- How to populate databases?
- Physiological data
- Simulation data
- What metadata are necessary?
- How to develop standards, visualization modes, and use examples for analysis routines?
- What tools would be useful to community?
- Individualized vs. central databases
Focus on Physiological Data
- Electrophysiology major gap in BIRN and similar shared resources
- Unifying feature is anatomy
- Spatial domain
- Electrophysiology deals with time series data
- Patch clamp, field potentials, EEG: continuous voltage as function of time
- Spike trains: list of event times
- Both require time linked event markers for stimuli and/or behavior for functional meaning
Diversity of Physiological Data Sources
- Different species
- Specialization for specific modalities
- Whole animals, slices, cell cultures
- Vertebrate or invertebrate
- Scales: channels, neurons, multiunit, ensembles
- Varied measurement techniques
- Non-standardized equipment
- Multiple manufacturers
- Custom made for individual labs
What kind of physiological data to share?
- Multiple examples of responses to given stimulation protocols from individual labs and groups of labs
- Submitter decides what levels of data to contribute
- Raw data (voltage records)
- Single trial data
- Averaged data (histograms, evoked potentials)
- Re-analysis needs of community are varied
Databases: top down and bottom up
- Neurodatabase.org
- Framework designed top down facilitating standardization
- Metadata formats informative but labor intensive for data entry
- Metadata format as prototype for organizing individual lab data
- Allow batch upload to database for streamlining input
- Neurosys
- Framework designed bottom up (lab notebook)
- Customizable and flexible for individual lab needs
- Reuse in individual experiments in lab
- Hard to transfer between labs and to central database
- Interoperability needed for data transfer from electronic notebook to central repository
Recommendations: Improvements to databases
- Better search mechanisms
- Stimulus characteristics
- Response characteristics
- Better descriptors for stimulus and behavioral protocols
- Development of stimulus description language
- Development of behavioral protocol descriptors
- Synergy with functional imaging databases and protocols
- Descriptors of neuronal responses as metadata
- Temporal firing patterns: phasic, tonic, bursty, rhythmic
- Receptive field classes
Recommendations: Development of modular analysis tools
- Tool repository
- Format conversion
- Spike sorting
- Descriptive statistics
- Single and multivariate time series
- Visualization
- Documentation
- Tutorial
- Examples
- Technical
- Shared use with functional imaging and other modalities
- Spectral analysis
- Source separation (PCA, ICA)
Recommendations: Physiological Simulations
- Facilitate comparison of simulated and experimental data (a two-way street)
- Identify and re-use key algorithms, routines, and methods useful across analyses
- Provide a test bed for new analysis techniques
- Metadata for computational models need same level of attention as experimental metadata
Reasons to submit data
- Submission increases visibility of research data
- Reanalysis with new tools and points of view
- Data immortality
- Recognition in grant progress report as shared resource
- Quality assurance
- Development of common descriptors across domains
- Which protocols yield most useful and reliable data?
- Further development of modular analysis tools
