Single unit analisis

Script to process single unit data and quantify activity (e.g. spontaneous peaks and correlations) on arbitrary epoch definitions.

Installation

• Download and install MATLAB.
• Download and extract source files to Documents/MATLAB
• For analyses involving deconvolution of CA²⁺ data, download and implement such library in the code. For example, download OASIS library, run oasis_setup.m, and use one of the examples under examples.

Usage

• Run setup.m to add FPA dependencies to the MATLAB search path.
• Create a copy of the example script.
• Change the configuration variables and the data loader according to your data acquisition system.
• Define epochs.
• Change settings and thresholds to compute peaks and clusters.
• Click Run.

Epochs consists of an array of timestamps indicating start and stop times (in seconds) of a given condition {'epoch1', [start1, end1, start2, end2, ...], 'epoch2', ...}. For example:

  epochs = {'A', [100, 200], 'B', [200, 500], 'C', [500, 600, 600, 700]};

means there are three epochs called A, B, and C:

A: 100 to 200
B: 200 to 500
C: 500 to 600 then 600 to 700

Documentation

msa = MSA(data, configuration);

Normalize, filter, and detect peaks of spontaneous activity in user defined epochs.

Each column in data corresponds to activity from individual cells.

configuration is a struct with the following fields (defaults are used for missing fields):

• conditionEpochs - Epochs for different conditions: {'epoch1', [start1, end1, start2, end2, ...], 'epoch2', ...}
• thresholdEpochs - Epochs to include for peak threshold calculation.
• events - Event-triggered data; times at which a type of event occurs.
• resamplingFrequency - Resampling frequency (Hz).
• lowpassFrequency - Lowest frequency permitted in normalized signal.
• peakSeparation - Minimum time separation between two peaks.

Processing steps

• Resample signal and reference to a given frequency.
• Baseline correction modeled as an exponential decay of the low-pass filtered data (optionally using airPLS).
• Correct for motion artifacts by subtracting reference to signal, after a polynomial fit (optional).
• Remove fast oscillations with a low-pass filter.
• Normalize data as df/f or z-score according to settings.
• Find peaks of spontaneous activity in low-pass filtered data.

Normalization is calculated as (f - f0) / f1 where f0 and f1 can be data provided by the user or calculated using given functions:

See examples and source code for detailed analysis steps and default parameters. Units for time and frequency are seconds and hertz respectively.

License

© 2019 Leonardo Molina

This project is licensed under the GNU GPLv3 License.