Transcript neuroelectro_incf_2014

NeuroElectro.org
A window to the world’s neurophysiology data
Shreejoy Tripathy
University of British Columbia, Canada
Email: [email protected]
Twitter: @neuronJoy
Main Idea
• Given that there is an extensive neuron
electrophysiology literature, what can we
learn by compiling it?
PubMed search: neuron AND
(electrophysiology OR biophysical OR
neurophysiology)
>45K articles
Electrophysiology literature is
notoriously heterogeneous
Electrophysiology literature is
notoriously heterogeneous
Input resistance
measurement differences
NeuroElectro overall methodology
Semi-automated text-mining overview
• Identify within data tables:
– Neuron types (from
NeuroLex.org)
– Biophysical properties (in
normotypic conditions)
– Biophysical data values
“Experiments were conducted in acutely
prepared brain slices of 24- to 28-day-old (65–
120 g) male Wistar rats.”
• Experimental conditions
defined within methods
sections
• Text-mined data is then
checked by experts
6
Tripathy et al, 2014
NeuroElectro.org web interface
Code at github.com/neuroelectro
Data at neuroelectro.org/api
Database statistics
• Currently 100 neuron types, >300 articles
Extensive variability among
NeuroElectro data
Resting membrane potential
Input resistance
Tripathy et al, in revision
MΩ
mV
Netzebrand et al, 1999
9
Accounting for differences in
experimental conditions
• Explain variability in
electrophysiological data
through influence of
experimental conditions:
–
–
–
–
–
–
species/strain
electrode type
animal age,
recording temperature
in vitro/in vivo/cell culture
junction potential
Electrode type
10
Tripathy et al, in revision
Tripathy et al, in revision
11
Neuron clustering on basis of
electrophysiology
Whole-genome correlation of gene
expression and electro-diversity
Patterns of gene
expression
Systematic
variation among
neuron types
Electrophysiological
phenotypes
20,000 genes
Tripathy et al, in revision/in progress
12
Making hypotheses on electrophysiology gene expression relationships
• Explaining electrophysiological phenotypes in terms
of underlying gene expression (and vice versa)
Future directions
• Continuing to expand NeuroElectro
– More neuron types
– More domains
• Synaptic plasticity
• Continuing to demonstrate the value of data
integration
– How can we move to a situation where
experimentalists are willingly sharing their data?
Acknowledgements
• Pavlidis Lab @ UBC
• Urban Lab @ CMU
• Gerkin Lab @ ASU
Shreejoy Tripathy
Email: [email protected]
Twitter: @neuronJoy
URL: neuroelectro.org
Code: github.com/neuroelectro
15
Mapping neuron electrophysiology to
gene expression
Neuron type
resolution
Cell layer
resolution
20,000 genes
Neocortex layer
5/6
Neocortex L5/6
pyramidal cell
Neuron type to cell layer mapping is
approximate. Will be improved in future
iterations with high resolution data.
Neocortex
Neocortex
basket cell
16
Finding genes most correlated with
electrophysiological diversity
Assessing predictive power between
gene expression and electrophysiology