hochbaumgroup / hnccorr Goto Github PK

Hi,

Thank you for this package! Does this assume motions are already been corrected?

Bests
Abed

Hi,

First thank you for this package! When I ran the given example, I encountered the error "Negative capacity for lambda equal to -0.000000. Set roundNegativeCapacity to 1 if the value should be rounded to 0." I simply copy and paste the code here and run.

Hello,
thanks for writing this library. i am a newcomer in Ca imaging analysis. I used your library on 1 .avi file after after building the HNCCorr Movie() with the .avi file's data. But no cells are identified when HCCCorr.segment(movie) is called. What are some possible issues with the data or code leading to this?

Hi, I have a question regarding parameters s.percRows and s.percCols.
In function approximatePCA, the full correlation matrix (which should be symmetric) is downsampled to nRows x nCols (line 38), and then PCA analysis (line 41) is carried out with the each column treated as a variable (according to MATLAB document), and the resultant top nComps PCA coefficient vectors are used to get PCA scores for all pixels (line 45-47).
In this sense, each row of the downsampled correlation matrix is a "feature vector" of a corresponding pixel as mentioned in the eNeuro paper, or equivalently, pixels corresponding to those columns are the smaller-size reference set. So we should have nCols < nRows.

However, in the class definition file Solution.m, we see:
obj.percRows = 0.01;
obj.percCols = 1;
This seems to contradict the analysis above. Have I misundertood or missed something in the code?
Thank you!