Coverage threshold on blast results

Question

[l.willianpacheco]

Hi folks

I'm working with two plant species, both eudicots, and I searching for best hit on Arabidopsis thaliana. Traditionally we applied an e-value and similarity threshold to transfer information from Arabidopsis to another species, but in relation to coverage.

What would be a good threshold to coverage?

My doubt is due to the possibility to find a small block with high similarity percentage and this could produce a false positive.

Thanks! Willian.