Many NAs on train_labels.csv

Hi. The file train_labels.csv has 267 columns (and one aditional column for cell_id as well), but ~98% of the columns are NA. Is this right?

Can we consider NA as 0? Plotting the distribution we see many zeros.