[Biopython] Normalizing paired distance in Treeconstruction

[Jordan Willis]

Hi,

I have a question about the pairwise distance calculation in the DistanceCalculator under Phylo.TreeConstruction.

It pertains to the normalization.

Line 456.

 return 1 - (score * 1.0 / max_score)


I understand that the goal here is to have a distance between 0 and 1. But couldn’t this be greater than 1 if say score was negative as is allowed in PAM and BLOSUM lookup matrices. Perhaps the entire distance matrix should be computed, then a minimum and maximum derived, and then normalized between 0 and 1 rather than do it on a pairwise basis. What do you think?

Jordan
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.open-bio.org/pipermail/biopython/attachments/20151013/25c4a685/attachment.html>