Interpretation of a dissimilarity score of zero

[kevinpryan]

Thank you for this great tool!

I am incorporating the dissimilarity score into a set of filters for potential neoantigens and am wondering how to handle dissimilarity scores that are precisely zero.

It is my understanding that currently, a dissimilarity score of zero can mean one of two things:

1. No self-proteome hits are found by blast, see line 309 of antigen.garnish_run.R

  if (all(file.info(blastdt)$size == 0)) {
    message("No self-proteome matches found by blast.")
    return(data.table::data.table(nmer = v, dissimilarity = 0))
  }

2. The peptide or one of its n-mers has at least one good match to the self-proteome (a high Smith-Waterman alignment score), leading to a dissimilarity of zero.

These two scenarios can mean two very different things. Scenario 1 may actually be due to high dissimilarity, and so if we are to filter by dissimilarity, we could lose potentially immunogenic neoantigens.

I would appreciate your advice on this.