I'm working on a piece of software that does comparative genomic analysis; and I found out in homology methods for functional annotation, it's preferable to pick the high scoring homolog from a distant sequence (i.e. not closely related to the query sequence in the phylogenetic tree).

My question is: why? and also could NCBI taxid for each taxon be used for approximating this distance? if not, then is there a simple way of making this binary decision (close, not close) when given two sequences.


1 Answer 1


No, because it is really not a binary decision. There are attempts to associate percent pairwise difference with taxonomic ranks (particularly species), but this is problematic because every independent lineage can vary according to its own rate. You could use sequences from different families, or orders, or even phyla, but these two can depend on whether specialists in the field are lumpers, or splitters. For example, all ants are in the same family, and Formicidae is over 100 million years old, while birds that may have diverged from a common ancestor within the last 10 million years are in a different family. You can use pairwise distance OR higher-level taxonomy as a proxy. But it would be crude. Why not program your software to make hard decisions about homology, and then find the closest most distant homologue in terms of pairwise differences?

  • $\begingroup$ are you referring to these methods: en.wikipedia.org/wiki/Distance_matrices_in_phylogeny $\endgroup$
    – 7kemZmani
    Commented Nov 15, 2017 at 3:27
  • $\begingroup$ and also what about the reason for picking a homolog that is not closely related to query sequence? is the sequence taxid useful at all? $\endgroup$
    – 7kemZmani
    Commented Nov 15, 2017 at 3:31
  • 2
    $\begingroup$ Yes, distance can be calculated simply by counting the differences between pairs, and then sometimes correcting those distances with various models that assume some of these changes will be masked by multiple changes at the same sites. The taxid tells about classification. Picking a distant homolog far from the query gives you more information about how much a gene can change over time. $\endgroup$
    – Karl Kjer
    Commented Nov 17, 2017 at 11:09

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .