This notebook processes analog odor-delivery data recorded from a HARP olfactometer. The pipeline includes:
The analog signal is resampled, smoothed, and differentiated. Peaks in the derivative mark pulse starts (local maxima) and pulse ends (local minima).
For each detected pulse, the code refines the true start/end using quantiles and separates the signal into pre-pulse, plateau, and post-pulse regions. It computes:
- baseline and plateau statistics
- linear regression slopes
- Pulse “squareness” -> Area under the curve (AUC) based
- Delay from the trigger origin (odor channel vs. end-valve)
Pulses are grouped by channel and end-valve origin (boolean that is true if the pulse has been triggered by the endvalve or False if it has been trggered by a change in the odor channel). Aggregated metrics (mean, std_dev, percentile errors) are computed for each group. Results can be exported to CSV.
