There are surprisingly many transferrable skills between reviewing code and reviewing a scientific paper, and in particular the advice on review tone in the excellent piece http://brson.github.io/2017/04/05/minimally-nice-maintainer (about code reviews) would apply equally to paper reviews.