Description of the data and file structure

Dataset overview

This repository contains haemodynamic and electrophysiological data supporting the publication:

“Hemodynamic and neuronal contributions to low-frequency vascular oscillations in a preclinical model of Alzheimer’s disease.”

The data were acquired from anaesthetised male J20 Alzheimer’s disease (AD) mice and wild-type (WT) controls aged 9–12 months, using 2-dimensional optical imaging spectroscopy (2D-OIS) and simultaneous electrophysiological recordings where applicable.

Abbreviations

• AD – Alzheimer’s disease
• WT – Wild-type
• 2D-OIS – Two-dimensional optical imaging spectroscopy
• LFO – Low-frequency oscillation (0.06–0.2 Hz)
• HbO – Oxyhaemoglobin
• HbR – Deoxyhaemoglobin
• HbT – Total haemoglobin
• Δ[Hb] – Relative change in haemoglobin concentration
• MUA – Multi-unit activity
• LFP – Local field potential
• FFT – Fast Fourier Transform
• ANOVA – Analysis of variance

Experimental summary

Objective:

To characterise low-frequency oscillations (LFOs; 0.06–0.2 Hz) in cerebral haemodynamics, assess their modulation by inspired gas challenges, and determine their relationship to underlying neuronal activity.

Signals recorded:

• Oxyhaemoglobin concentration (HbO)
• Deoxyhaemoglobin concentration (HbR)
• Total haemoglobin concentration (HbT)
• Multi-unit activity (MUA), where applicable
• Local field potential (LFP), where applicable

File Formats:

• .mat – MATLAB data files
• .m – MATLAB function/script files
• .mlx – MATLAB Live Scripts
• .R – R statistical scripts
• .csv – Comma-separated value data files
• .xlsx – Excel data files

MATLAB data files:

Raw haemodynamic data from the experiments are provided in MATLAB (.mat) format and contain measurements of changes in oxyhaemoglobin (HbO), deoxyhaemoglobin (HbR), and total haemoglobin (HbT) obtained using 2-dimensional optical imaging spectroscopy (2D-OIS). HbO, HbR, and HbT represent relative changes in haemoglobin concentration (Δ[Hb]) and are expressed in micromolar (µM). For animals with an implanted electrode, electrophysiological recordings are also included. Multi-unit activity (MUA) represents multi-unit spike activity recorded from the somatosensory cortex (Volts), and local field potential (LFP) recordings reflect summed synaptic activity from the somatosensory cortex (Volts).

Raw/Primary Data files:

• mouse_data_for_shannon_n49.mat
• new_data_for_shannon_osman.mat

Data structure:

A 3-dimensional matrix of size:

[sessions x regions x data points]

4 regions (1 = Whisker barrel cortex, 2 = Artery, 3 = Vein, 4 = Parenchyma)

6000 data points in the experiment (equivalent to a length of 750 seconds)

Group information:

N of 49 (mouse_data_for_shannon_n49.mat)

• 1-19 = WT
• 20-24 = WT Neural
• 25-44 = J20-AD
• 45-49 = J20-AD Neural

N of 24 (new_data_for_shannon_osman.mat)

• 1-5 = WT
• 6-11 = J20-AD
• 12-16 = WT Neural
• 17-24 = J20-AD Neural

Data split by inspired gas condition for analysis:

• mouse_data_for_shannon_n49.mat - data from Paul Sharp's research, oxygen condition
• new_data_for_shannon_osman.mat - data from Osman Shabir’s research, oxygen condition
• mouse_data_for_shannon_n49_air_experiment.mat - data from Paul Sharp's research, air condition
• new_data_for_shannon_osman_air.mat - data from Osman Shabir’s research, air condition

Data for acute sessions (neural):

Data files:

• Osman13_air_to_oxy_transition_ts_data.mat -HbT, LFP, and MUA data for mice with an electrode from Osman Shabir’s research (air to oxygen condition)
• Osman13_oxy_to_air_transition_ts_data.mat - HbT, LFP, and MUA data for mice with electrode from Osman Shabir’s research (oxygen to air condition)
• Paul_air_to_oxy_transition_ts_data_n10.mat - HbT, LFP, and MUA data for mice with electrode from Paul Sharp’s research (air to oxygen condition)
• Paul_oxy_to_air_transition_ts_data_n10.mat -HbT, LFP, and MUA data for mice with an electrode from Paul Sharp’s research (oxygen to air condition)

Data description:

These data files contain hemodynamic and electrophysiology data from experiments 3 and 6. The file is presented in a 13 x frames, or 10 x frames matrix, corresponding to:

23 (13 + 10) sessions

HbT: averaged from the arterial region; MUA and LFP: from the somatosensory cortex

6000/4581500/763482 data points in the experiment (equivalent to a length of 750 seconds) for HbT, LFP and MUA.

Downsampled data files:

• osman_a2o.mat - contains LFP, HbT, and MUA data (downsampled) for mice with an electrode from Osman Shabir’s research (air to oxygen condition)
• osman_o2a.mat - contains LFP, HbT, and MUA data (downsampled) for mice with an electrode from Osman Shabir’s research (oxygen to air condition)
• paul_a2o.mat - contains LFP, HbT, and MUA data (downsampled) for mice with an electrode from Paul Sharp’s research (air to oxygen condition)
• paul_o2a.mat - contains LFP, HbT, and MUA data (downsampled) for mice with electrode from Paul Sharp’s research (oxygen to air condition)

Data description:

Similar structure to the data file above. Downsampled for further analysis. 

Statistical Analysis and Figures:

Chronic Sessions (HbT) - Linear Mixed Model

Analysis Code (.R file): LFO_Data_Analysis_HbT.R

LFO_Data_Analysis_HbT.R contains code used to run the linear mixed model and also contains code to create Figure 4

mice_vaso_avefft_figure_code.mlx

Preprocessing and power analysis

Data sheet (Excel file): LFO_HbT_Data_Chronic.xlsx

LFO_HbT_Data_Chronic.xlsx contains summed LFO power from each session, in addition to group and animal information

Acute Sessions (HbT) - ANOVA

Analysis Code (.R file): LFO_Data_Analysis_HbT.R

LFO_Data_Analysis_HbT.R contains code used to run the ANOVA and also contains code to create the acute data plots shown in Figure 5

Data sheet (Excel file): LFO_HbT_Data_Acute.xlsx

LFO_HbT_Data_Chronic.xlsx contains summed LFO power from each session, with corresponding group and breathing condition information.

The MATLAB file mice_vaso_avefft_figure_data.mat includes all the data required to create the time series and FFT plots featured in Figure 3. The figure can be created using the MATLAB code Figure3code.m

Kernel Analysis

final_code.mlx (MATLAB Live Script): This script performs kernel regression between MUA and HbT. It was used to generate Figures 2, 6a, 6(b1), 6(b2), and 8d.

Statistical Analysis:

Data Files:

Kernel shape & prediction: anova_input_final.csv

Contains kernel prediction and kernel shape parameter values used as input for the ANOVA statistical model. 

LFP Before and after experiment: LFP_anova_input_begain.csv and LFP_anova_input_after.csv contain LFP power across different bands before and after the experiment. LFP_anova_input.csv used as input data for ANOVA.

MUA Power across conditions: mua_power_log.csv and mua_power_nolog_final.csv 

Kernel shape parameters: kernel_spec.mat and kernel_spec_noarti.mat 

Code Files:

Kernel shape & prediction: ANOVA_kernel.R

• Performs mixed ANOVA and assumption tests.
• Used to generate Figures 6(c1)–6(c3).

ANOVA for low-frequency MUA power: ANOVA_kernel.R

• Performs mixed ANOVA and assumption tests.
• Output used to generate Figures 5c and 5d.

MUA_figures_code.R creates the graphs for 5c and 5d using the data file mua_log.xlsx

LFP ANOVA: LFP_linearMix.R (Note: This performs mixed ANOVA, not linear mixed modelling)

• Used for mixed ANOVA and assumption testing.
• Generates Figures 8b and 8c.

final_code.mlx: Preprocessing, Main analysis, and visualisation for publication

• Generates kernel figures and Figure 8d

Help Function:

• ComputeAvPowerSpectrum.m: compute the average power spectrum, by slicing the input to M windows (half overlapping), computing the spectrum for each, and averaging.
• compute_band_power.m: Calculates the sum of power (electrophysiology) in different frequency bands.
• kernel_slope.m: Calculate kernel shape parameters.
• Ldeconvs.m, Neural_vascular.m, Neural_vascular1.m: Deconvolution function.
• LFP_band_kernel.m: Calculates the kernel between LFP in each band and HbT.
• mouse_kernel.m, mouse_kernel_LFP.m, mouse_kernel_mua.m, my_plot_kernel.m: For visualisation during analysis.
• showfft.m: MATLAB function, displays the amplitude spectrum of an FFT
• mouse_regress.m, my_ANOVA.m, my_mixed_anova_assumtion.m: A quick linear regression/ ANOVA test.

R code is used for all statistical analyses in the publication.