NWB Workshops and Hackathons

Back to the projects list

Stimulus representation and parametrization in NWB

Key Investigators

• Pamela Baker
• Shreejoy Tripathy
• Lydia Ng

Project Description

The problem: labs performing icephys are often delivering the same (or very similar) types of stimuli (e.g., long steps, ramps, chirps, etc). Despite the stereotyped nature of these stimuli, the NWB schema currently does not support a standardized stimulus representation, leaving it up to the user to determine how to describe the stimulus protocol and what stimulus metadata to include. This lack of standard representation leads to difficulty in identifying similar protocols across datasets and development of shared analysis scripts for quantitative feature extraction.

In addition, there is often considerable redundancy in the stimulus set that defines a single experimental protocol because the stimuli typically differ by only a single parameter (e.g. amplitude, latency, frequency) that is varied over some range of pre-selected values to generate a tuning curve or compute a metric. These variable parameters should be made explicit in the stimulus representation. Parametrization also allows stimuli to be represented compactly, and is practically a necessity for more complex types that can not be represented as 1D time series.

This work applies not only to patch clamp experiments but will also be directly applicable to other 1D time-varying stimulus types, including extracellular electrical stimulation and optogenetic stimulation protocols.

Objectives

1. Define a canonical set of 1D stimulus waveforms typically used in IC ephys experiments, and the parameters needed to describe each stimulus type (minimum to specify).

2. Develop a representation in NWB that supports parametrized stimulus definition and exposes variable parameters for search/revealing experimental design.

3. Determine the minimum set of stimulus metadata that should be defined at each level of the experimental protocol (sweep/sequence/stimset/condition/etc) for comprehensive experimental descriptions.

4. Design desired visualizations for stimuli and stimulus protocols with widgets.

Approach and Plan

• For 1-D waveform types we already have e.g. the WaveformBuilder in MIES that defines a base set of stimulus types, with associated parameters, as a starting point. It appears that the Channelpedia stimuli can be constructed from the same set of elements as well.

• Stimulus representation also dovetails into the issue of overall experimental protocol representation, so we will also end up touching on issues that relate to provenance. I would like to discuss some ideas for how stimulus parameterization and metadata representation (objectives 3 and 4) could be integrated with the IC ephys hierarchical tables extension.

Progress and Next Steps

Materials

Background and References

• Luke Campagnola had a previous Hackathon project that made a good start on some of these issues, specifically the issue of stimulus atomization/composition (see [here:] (https://github.com/NeurodataWithoutBorders/nwb_hackathons/tree/master/HCK04_2018_Seattle/Projects/StimulusMetadata)).