lengning / oefinder Goto Github PK

Hi,

First of all, thank you for pointing out the ordering effect of capture sites. As a result, we are now beginning to explore this issue in our data. I recently encountered an error message while running OEFinder. I wonder if you can help me to decipher this message...

Warning in summary.lm(aa) :
essentially perfect fit: summary may be unreliable

Best,
Joyce Hsiao

Postdoctoral fellow at University of Chicago

Currently, I'm trying to test out the Shiny application that's included with the repo. My data is this:

count_table.zip

This is an example dataset that we've included with our own workflow project that's hosted on GitHub. Using the default settings, the matrix ends up with a bunch of NAs. Why? Because of this expression chain within the function MedianNorm from EBSeq:

geomeans <- exp(rowMeans(log(Data)))
    out <- apply(Data, 2, function(cnts) median((cnts/geomeans)[geomeans > 
      0]))

Noticeably, if the geomeans end up being zero, things end up being NA.