NA in the coefficient

[Huimin721]

Dear team,

Thanks for developing the package. I was running the function on a diabetes dataset:

data = data.frame(Group = Group,ID = ID,imageID = imageID,x=x,y=y,cellType=cellType)

res = spicyR::spicy(
    cells=data,
    condition = 'Group',
    subject = 'ID',
    cores = 80)

This is how the dataset look like:

str(data)
'data.frame': 1292937 obs. of 6 variables:
$ Group : chr "Onset" "Onset" "Onset" "Onset" ...
$ ID : chr "6362" "6362" "6362" "6362" ...
$ imageID : chr "A01" "A01" "A01" "A01" ...
$ x : num 52.5 128 135.3 449.5 458.4 ...
$ y : num 0.667 0.895 0.429 1.158 1.1 ...
$ cellType: chr "acinar" "acinar" "acinar" "acinar" ...

The function will run but get a one column coefficient full of NA, and my package version is 1.19.
Could I ask why this happens? Should I set the characters to be factors?
Thanks so much!

[shreyarajeshrao]

Hi @Huimin721,

Thanks for bringing up the issue! It looks like celltype being a character might be causing problems with the model fit, so converting it to a factor might fix the issue. If you have very few cells per image, spicyR might have difficulty generating a window, so you could try window = "square" instead of the default window = "convex". If neither of these work, could you share a subset of your data with us so we can find the issue?

Thank you!
Shreya