Matlab scripts for joint behavioral and neural analysis of Intan recordings

This Matlab script can be used for analyzing how neural activity correlates behavior including lick, facial activity, speed, and up to four external manipulations. All Matlab scripts were written in Matlab 2022b with all official packages and software.

To see what results this code can achieve, please see our research paper bioRxiv 2023.07.11.548619; doi: https://doi.org/10.1101/2023.07.11.548619

To install Matlab, please view https://www.mathworks.com.

To install Jupyter Lab, please view https://jupyter.org/install.

Prerequired codes and software: 'read_Intan_RHD2000_file.m', 'upsample2x.m' available from https://intantech.com/downloads.html?tabSelect=Software
                                TSPE toolbox https://github.com/biomemsLAB/TSPE
                                Bayesian adaptive kernel smoother https://github.com/nurahmadi/BAKS
                                Spike sorter swindale.ecc.ubc.ca/home-page/software
                                Stimulus-associated spike latency test http://kepecslab.cshl.edu/software

All installation can be done in a day.

The behavioral set up for mouse can refer https://github.com/ywang2822/Multi_Lick_ports_behavioral_setup

To jointly analyze behavioral and neural data, we first extract these data separately.

The behavioral metrics include lick signal, images with time evolving (facial activity), speed (locomotor activity), and stimulation onsets.
for individual trial: 'Lick_BehavDat.m' 
for combined trial: 'lick_BehavDat_stack.m'
for analysis of all trials: 'Data_Stac_Ave.m'
helping function: 'list_lick.m' (input trial name in this function)

The neural activity data is specifically spiking data.
SpikeSoter: save as both csv and plx file
To generate and align spike raster data from csv file: 'Lick_SpikDat.m' (used for both individual and combined trial)
To identify opto-tag neural activity channel: 'Cell_ID.m'

Second, we categorize cells based on whether their activiy can represent movement phases, valence, or lick command.
For valence: 'evaluate_valenced_selectivity_byBAKS.m'
             helping codes: 'evaluate_valenced_selectivity_byBAKS_wq.m','evaluate_valenced_selectivity_byBAKS_wsq.m'
For movement phases: 'evaluate_cue_selectivity_byBAKS.m' Please note 'DO' denotes initial-before lick onset, 'LO' denotes initial-after lick onset, 'quinine' denotes terminal.
The generated data should be saved in such order 'Neuron_index'>'valence'>'wsq'(session)>'IC'(brain region)
To calculate the number of these categorized neural representations: 'evaluate_number_selected_cell.m'

Third, we calculate how these different categorized neural representations connect with each other during behavior.
evaluate connectivity among and in each category: 'evaluate_properties_of_selected_cell.m'
(this code can also evaluate trendline slope and trajectories angles of water lick trials)
evaluate connectivity between optotag and optoexcited neurons: 'evaluate_connectivity_optoexi.m'

Fourth, we evaluate if the neural activity of certain group of cells (e.g.projection defined or genetically defined) can predict behavior.
For large-scale window (movement phases) and small-scale window (valence): 'decode_liquid_type_singlemouse.m'
For decoding all lick trials (valence only): 'decode_all_lick_trials.m'
For decoding facial activity: 'behavioral_modeling.m'

Fifth, we display the mean neural activity in certain brain regions
To evaluate mean firing rate: 'mean_neural_dynamics.m'
To evaluate whether there is a linear correlation between mean firing rate and behavioral data: 'mean_neural_dynamics_singletrial.m'

Finally, we simulate the behavioral output by using spiking neural network.
To generate spike data: 'Network_modeling.ipynb'
To generate behavioral data: 'network_moedeling.m'

One data files folder is provided in 'Example of original data file folder'.

For any question, please contact robertwangyihan@outlook.com