This toolbox provides a collection of MATLAB functions designed to facilitate the analysis of neural and behavioral data. It covers estimating tuning curves, fitting generalized linear models, decoding spike train data, detecting cell assemblies, and performing time-frequency analysis. It was written in August 2023 for the Summer school "Advanced computational analysis for behavioral and neurophysiological recordings." It is associated to a set of tutorials which cover the aforementioned analyses step-by-step.
- Load functions
- Mapping neural responses along one or two variables
- Generalized Linear Models (GLMs)
- Bayesian decoding
- Cell Assembly Detection
- Cross-Correlation Analysis
- Time-Frequency Analysis
The Load folder contains functions for loading and preprocessing data from the example dataset using functions like SetLoadParams, LoaddataNav, LoaddataSpk, and LoaddataLfp. These functions should be customized to your specific datasets while keeping the main logic intact. They return structures containing different types of data: LoaddataNav should return a structure (called Nav here) containing data about experimental conditions and explanatory variables, such as the appearance of a stimulus or the position of the animal; LoaddataSpk should returns a structure (called Spk here) with spike counts sampled at the sampling rate of data stored in Nav together with spike times and information about the recorded cells; LoaddataLfp should return a structure (called Lfp here) that contains signals that needs to be sampled at a higher sampling rate than data in Nav, for instance Lfp signals for which you would like to investigate the frequency content in the gamma range.
For more information and usage examples, refer to the related documentation and Tutorial1.
The Maps section includes functions for estimating tuning curves along up to two variables, using variables stored in the structure Nav mentioned above and a set of responses, typically corresponding to spike trains stored in Spk. SetMapsParams defines options and MapsAnalysis computes tuning curves and their related statistics.
For more information and usage examples, refer to the related documentation and Tutorial2.1.
In the GLMs section, you'll find functions to estimate GLMs with up to two variables and evaluate their significance. The predictors of the GLMs are variables selected from the structure Nav which are used to fit a set of responses, usually corresponding to spike trains stored in Spk. Use SetGLMsParams to define options and GLMsAnalysis to estimate tuning curves using a Poisson GLM model.
Learn more from the function documentation, Tutorial2.2 and Tutorial_handson.
The Decoding section provides functions for decoding up to two variables from neural spike trains using a Bayesian approach. Again, this is using variables loaded into the structure Nav and a set of responses, corresponding to spike trains stored in Spk. Use SetDecParams to define parameters and DecodingAnalysis to estimate decoded positions and errors.
For usage examples and more information, refer to the function documentation, Tutorial3.1 and Tutorial_handson.
The Patterns section contains functions for detecting cell assemblies from neural spike response data using independent component analysis (ICA). It uses variables loaded into the structure Nav and detect coincidental activity among a set of responses, typically corresponding to spike trains stored in Spk. Use SetPattParams to define parameters and PatternAnalysis to identify cell assemblies.
Detailed usage examples can be found in the function documentation, Tutorial3.2 and Tutorial_handson.
In the Corr section, you will find functions to estimate cross-correlations between pairs of neural responses and extract noise/trial-by-trial correlations using a shuffling to estimate signal correlations related to some varaibles stored in Nav. Use SetCrossCorrParams to set parameters and CrossCorrAnalysis to compute cross-correlation results.
For detailed information, usage examples, and tutorials, see the function documentation, Tutorial3.3 and Tutorial_handson.
In the TFMaps folder, you'll find functions to perform time-frequency analysis of continuous data and compute power and coherence maps along one or wo explanatory variables stored in the Nav structure. Use SetTFMapsParams to define options and TFMapsAnalysis to compute maps from wavelet transform and wavelet coherence between a set of recording channels.
For detailed usage examples, refer to the function documentation, Tutorial4.1 and Tutorial_handson.
This project is licensed under the MIT License.
This toolbox was developed by J. Fournier and Tulio Fernandez de Almeida in August 2023 for the Summer school "Advanced computational analysis for behavioral and neurophysiological recordings."