Using the cell extraction output when extracting cells with cnmf-e

[luciedescamps]

Hi Biafra,

Some people in the lab are getting started with new imaging projects and have been using the calciumImagingAnalysis package with great success to preprocess movies and extract cells.
They've been using cnmf-e and I am a bit unsure about what should be used from the cnmfeAnalysisOutput file for further data analysis.
For example when I extract cells from movies for my project with Extract, I then use extractAnalysisOutput.traces. But cnmfeAnalysisOutput has extractedSignals, extractedSignalsEst, and extractedPeaks.

What is exactly the difference between these 3, and do you have any advice on what is best to use when one wants to plot rate maps for example?

Thanks!
Lucie

[bahanonu]

@lalde
Great to hear CIAtah is working out for you-all! Let me know if there's anything they'd like improved or features added.

For CNMF-e, I normally use either extractedSignals or extractedSignalsEst combined with ciatah's computeSignalPeaks function to detect the location of transients when I use those instead of the inferred dF/F. See below figure.

• extractedSignals is CNMF-e's denoised, model activity trace constrained by calcium dynamics. This is usually a version with relatively little noise and can be used once you verify it matches extractedSignalsEst and activity in the movie (e.g. it is not getting too many false positives or negatives).
• extractedSignalsEst is CNMF-e's estimate of dF/F (scaled). This is often preferred to get a sense of the actual dynamics in the movies, I'll use esp. when initially looking at the data.
• extractedPeaks gives an estimate of spiking activity. I do not use this output as much for miniscope data since interpreting exact peak numbers can be difficult.

Let me know if you have any follow-up questions.

[luciedescamps]

Awesome - thanks a lot for the clarifications! Will let you know if any follow-up questions arise.

Lucie