Zero counts for ligands

[BioinformaNicks]

Hi,

Firstly, I really like the idea behind the manuscript, particularly the HLEP/HLDP formula.

I'm currently trying to replicate it for some other analyses, and I noticed that for this part:

We first counted the number of predicted ligands Ni of length l coming from protein i. The average number of presented ligands for each HLA allele was computed as <ρ = /<Li, where Li = length of protein – l is the corrected protein length and <·> denotes the average over the proteome.

zero counts for a protein were not included? Is there a particular reason for this?

I do notice you include a pseudocount in the actual code (to circumvent zero counts?):

mutate(odds = log2((count + 1) / (protein.len + 1 - ligand.len) / P0),
p.value = binom.test(count + 1, protein.len + 1 - ligand.len, 
p = P0, alternative = "two.sided")$p.value)

However, there are no zero counts in the data provided?