Skip to content

Things to keep in mind

Jump to bottom
Talia Lerner edited this page Oct 29, 2021 · 39 revisions

a. Please read all the instructions provided in with the GuPPy GUIs and in this Wiki. If you think the instructions are confusing, please send us your feedback! We are always happy to take your feedback into consideration to help us improve the accessibility of our tool.

b. Please follow the naming convention instructions provided in the Storenames GUI. It is very important to follow these instructions, otherwise, GuPPy will throw an error. GuPPy can take multiple channels for analysis so it is very important to follow the naming convention.

c. How does the filtering of different channels take place?

  • GuPPy uses a zero phase moving average linear digital filter to filter out the data.
  • The moving average filter requires a window to be set while using the filter. The default window GuPPy uses is 100 datapoints.
  • You may want to change this parameter based on the sampling rate of your data.
  • For example, when the sampling rate is ~1000 datapoints/second, a window of 100 datapoints seems ideal to use.
  • For data recorded at lower sampling rates like 20Hz (common for Neurophotometrics data), a window of 5 or 10 is more appropriate.

d. What happens in the analysis when the user is not using an isosbestic control channel (setting parameter Isosbestic Control Channel? = False)?

  • GuPPy will recognize that there is no isosbestic control channel.
  • It will create a control channel by fitting an exponential function to smoothed version of the signal channel and save that new information into the output folder.
  • All the other processes like artifacts removal and computation of ΔF/F will use this estimated control channel.

e. How the artifacts removal process is carried out?

  • The user is required to select chunks of data that do not have artifacts in the artifact removal process.
  • After those chunks are selected, computation of ΔF/F is carried out for each chunk separately, and finally, all the chunks are concatenated together.
  • z-score for the full ΔF/F trace is computed to get the final z-score of that trace.
  • GuPPy automatically keeps behavior timestamps properly aligned to the data when removing chunks.

f. How does the z-score computation work? What are my options for computing the z-score?

  • GuPPy calculates the z-score for the whole recording session.
  • GuPPy provides three different computational methods to compute the z-score.
  • These computational methods are as described in the table below.
Pros Cons Formula
standard z-score 1. Useful
for extracting
continuous values
for analysis
2. Widely used and easily
interpreted		 1. May not be as useful
for finding outliers as
alternatives
baseline z-score 1. Useful for finding
outliers
2. Continuous z-scores
can still be interpreted
in standard language		 1. Typically not appropriate
to satistically compare time
periods within single trials
modified z-score 1. Can be more effective
finding outliers in
smaller datasets.		 1. Not used as commonly as
others methods of
standardization
2. Difficult to interpret in
plain language
Clone this wiki locally