This repository contains the code that accompanies our data release Kyzar et al. 'Dataset of human-single neuron activity during a Sternberg working memory task'. The purpose of the code in this repository is to provide examples of how to use the released data. This dataset is formatted in the Neurodata Without Borders (NWB) format, which can easily be accessed from both MATLAB and Python as described here.
This code accompanies the following data descriptor:
- Kyzar M, Kamiński J, Brzezicka A, Reed CM, Chung JM, Mamelak AN, Rutishauser U. Dataset of human-single neuron activity during a Sternberg working memory task. Sci Data. 2024 Jan 18;11(1):89. doi: 10.1038/s41597-024-02943-8. PMID: 38238342; PMCID: PMC10796636. Link to Paper
The papers that describe the scientific results that are based on this dataset are:
- Kamiński, J. et al. Persistently active neurons in human medial frontal and medial temporal lobe support working memory. Nature Neuroscience 20, 590-601 (2017). https://doi.org:10.1038/nn.4509
- Kamiński, J., Brzezicka, A., Mamelak, A. N. & Rutishauser, U. Combined Phase-Rate Coding by Persistently Active Neurons as a Mechanism for Maintaining Multiple Items in Working Memory in Humans. Neuron 106, 256-264.e253 (2020). https://doi.org:10.1016/j.neuron.2020.01.032
Abstract of the paper:
We present a dataset of 1809 single neurons recorded from the human medial temporal lobe (amygdala and hippocampus) and medial frontal lobe (anterior cingulate cortex, pre-supplementary motor area, ventral medial prefrontal cortex) across 41 sessions from 21 patients that underwent seizure monitoring with depth electrodes. Subjects performed a screening task (907 neurons) to identify images for which highly selective cells were present. Subjects then performed a working memory task (902 neurons), in which they were sequentially presented with 1-3 images for which highly selective cells were present and, following a maintenance period, were asked if the probe was identical to one of the maintained images. This Neurodata Without Borders formatted dataset includes spike times, extracellular spike waveforms, stimuli presented, behavior, electrode locations, and subject demographics. As validation, we replicate previous findings on the selectivity of concept cells and their persistent activity during working memory maintenance. This large dataset of rare human single-neuron recordings and behavior enables the investigation of the neural mechanisms of working memory in humans.
This repository can be downloaded by entering the following commands:
cd $target_directory
git clone https://github.com/rutishauserlab/workingmem-release-NWB.git
Running the provided code and analyzing the dataset in MATLAB requires the download and initialization of MatNWB, a MATLAB interface for reading and writing NWB 2.x files. Instructions for how to download and initialize MatNWB have been listed on the project's public git repo. Further documentation for how to use MatNWB can be found here. MatNWB version 2.6.0.2 was used for the curation and analysis of this dataset.
The dataset is available in NWB format from the Dandi Archive, under Dandiset #469.
Dandi datasets are accessible through the Dandi command line interface (CLI). To install this Python client, use pip install dandi or conda install -c conda-forge dandi, depending on your Python environment setup.
After installing the Dandi CLI, use dandi download https://dandiarchive.org/dandiset/000469 to download the dataset.
NWB Files can additionally be loaded and analyzed using the PyNWB python package. Further documentation can be found here.
Validation of this dataset was performed using PyNWB (2.3.1) and PyNWB-dependent packages, such as nwbinspector (0.4.28) and dandi (0.55.1). The command lines used for each method are as follows:
- dandi:
dandi validate $target_directory - nwbinspector:
nwbinspector $target_directory - PyNWB:
$file_list = Get-ChildItem $target_directory -Filter *.nwb -Recurse | % { $_.FullName }; python -m pynwb.validate $file_list
All validators returned no errors in data formatting & best-use practices across all uploaded files.
The main script in this repo, NWB_SB_import_main.m, is designed to analyze the released dataset and to reproduce the figures & metrics noted in Kyzar et. al. 2023, Kaminski et. al. 2017, and Kaminski et. al. 2020. It can calculate several metrics related to behavior (reaction time, accuracy), spike sorting, and single-unit (SU) activity during the screening (SC) & Sternberg (SB) tasks.
-
Set Parameters: The first section of the script sets important parameters. The
taskFlagis used to specify which tasks are accessed and are defined as1(SCREENING),2(STERNBERG), or3(BOTH). TheimportRangeis the range of subject IDs for the dataset. For the current release, subject IDs have a range of1:21. -
Initialization and Pathing: The script then defines the directory paths for the code, the currently installed MatNWB package, and the dataset, and then adds them to the MATLAB path. If figures are generated, there is an additional option to add a custom save destination. Please ensure that the defined paths in the script are correct for your setup. This section also uses MatNWB's generateCore() function to initialize the NWB API if it has not been initialized already.
-
Import Datasets From Folder: The script will then import datasets from the given folder using the
NWB_importFromFolderfunction. Only files specified usingimportRangeandtaskFlagwill be loaded into the workspace. -
Extracting Single Units: Single unit information is extracted from the loaded NWB files for ease of indexing, using the
NWB_SB_extractUnitsfunction. If spike waveforms are not needed for analysis, theload_all_waveformsflag can be set to0to only extract the mean waveform. All future plots will use this mean waveform instead of a spike waveform pdf. -
Screening Analysis: This section is preceded by a parameters section, which allows for the control of various stages of the analysis and plotting process. For example, one can choose to plot figures for significant cells by setting
paramsSC.doPlot = 0or filter units being used for analysis by specifying a minimum firing rate thresholdparamsSC.rateFilter. To disable screening analysis of all cells entirely, setparamsSC.calcSelective = 0. -
Sternberg Analysis:: This section is also preceded by a parameters section, with additional controls included for the type of cell type to plot (
1: Concept,2: Maint,3: Probe,4: All). Similar to screening, analysis can be disabled by settingparamsSB.calcSelective = 0. -
Example Neurons: Additional sections have been added for each task type that optionally plots the example cells found in Kyzar et al 2023 & Kaminski et al 2017. To reduce load times for NWB files, set
importRange = [4 7 15 16 21]for screening andimportRange = [7 14 16]for Sternberg. -
Selectivy by Area: The script also calculates the proportion of selective cells by area and plots bar/pie charts comparing the screening and Sternberg tasks. This can be disabled by setting
plotAreas = 0. -
Spike Sorting Quality Metrics: This section plots spike sorting metrics for single units recorded in the Sternberg/screening tasks. These metrics include the percentage of inter-spike intervals (ISIs) that were less than 3 ms, mean firing rates for all units, coefficient of variation (CV2) values for all units, signal-to-noise ratio (SNR) of the peak of the mean waveform, mean SNR for all values in a unit’s mean waveform, pairwise projection distance between each unit in which multiple units were found on the same electrode, isolation distance (scaled to log 10 for ease of viewing) across all units for which this metric was defined.
Please make sure to thoroughly read the comments in the code to understand the functionality of each part. If you encounter any problems, please report them as issues in the repository.
- Michael Kyzar
- Ueli Rutishauser (Principal Investigator)
Acquisition of this dataset has been supported by the National Institute of Mental Health (R01MH110831) and the National Institute of Neurological Disorders and Stroke (U01NS117839, U01NS098961).
"workingmem-release-NWB" Copyright (c) 2023, Rutishauser Lab. All rights reserved.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
-
Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer.
-
Redistributions in binary form must reproduce the above copyright notice, this list of conditions and the following disclaimer in the documentation and/or other materials provided with the distribution.
-
Neither the name of the copyright holder nor the names of its contributors may be used to endorse or promote products derived from this software without specific prior written permission.
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.