Broad Goal - Identify differences in hippocampal activity during REM sleep as a function of whether a rat carries Alzheimer's disease related transgenes.
Approach - Implant rats with versus without the AD transgenes with silicon depth electrodes targeting the dorsal hippocampus and then collect 2 hour recordings during the 'light-phase' of the sleep cycle (when nocturnal rats should be sleeping) in a quiet isolated room. Using the recordings to analyze the activity during REM sleep. Analyses can start simple and become more elaborate as we go based on what we find.
Goal 1: Compare the frequency and duration of REM sleep epochs between TG and WT rats
Background
REM sleep is defined as moments of stillness when the hippocampal theta rhythm (8Hz) is two times larger than the hippocampal delta rhythm (2Hz)
Method:
Locate relevant data
Synchronize behavioral videos with brain activity recordings (ephys)
Score the behavior using DeepLabCut (DLC)
Import synchronized DLC behavior data and ephys recordings
Find moments of stillness
Threshold velocity to find low-velocity epochs
Find moments of high theta power
Threshold the ratio of theta to delta power to find these
Find moments of REM sleep
These are the epochs when the animal is still and the theta power is high
Compare frequency of REM epochs between WT and TG
Compute the mean frequency per animal
Compare the means across groups of animals with stats
Compare duration of REM epochs between WT and TG
Compute the mean duration per animal
Compare the means across groups of animals with stats
Skills:
Orange - Talk to the relevant people to find the data
Yellow - DLC: Use DLC on HPC to score videos (may involve building new CNN)
Green - Matlab: Update existing scripts to handle current data, may involve handing changes in data format / paths
Blue - Matlab: Using CMBHOME object, thresholding, saving results
Purple - Matlab: stats and aggregation. Will need to write your own using stats MATLAB tools
Goal 1.1 - Set up REM detection scripts using already imported data
General method -
Use Dylan's MS-Opto rat data already imported
Load saved CMBHOME object
Explore root.user_def
Goal 1.1.1 - Find high theta epochs
Method -
Load 1 channel of LFP
Find theta power in root object
plot theta power over full recording
use 'OverThresholdDetect' to find epochs when theta power is 2 standard deviations larger than the mean power
Bonus: find 'epochs' of high theta power, where epochs is a Nx2 matrix, with a row for each of the N detected epochs, each with the start index and the end index.
Goal 1.1.2 - Find low movement epochs
Method -
Using same data as already using...
Find velocity in the root object
plot velocity over full recording
use 'UnderThresholdDetect' to find epochs where movement speed is less than 8 cm/sec, store this in root.user_def, run cmbSave(root)
Plot this over the velocity to highlight epochs of stillness
Bonus: find 'epochs' of stillness, where epochs is a Nx2 matrix
Goal 1.1.3 - Find theta-delta ratio
Method -
Double check delta is not already computed by CMBHOME. If it is, hurray! Use it.
If not already computed by CMBHOME, compute the delta power over the full recording (use 'buttfilt.m' to find 1-4 Hz power)
plot delta power over full recording
compute ratio of theta and delta, plot over full recording
use 'OverThresholdDetect' to find epochs when theta-delta ratio is 2 or larger, store this in root.user_def, run cmbSave(root)
Bonus: find 'epochs' of high theta-delta power
Goal 1.1.3 - Putting it together
Find where rat is still and there is high theta-delta ratio - our surrogate for REM sleep detection!
Find 'epochs' of REM sleep. Store this in root.user_def, run cmbSave(root)