[Biodevelopers] Re: blast

[Michael Nuhn]

Hello, Pedro!

Thanks for your response.

I am aware of the heuristic to detect orthologs/paralogs.

In my case however, there is no sequence C in the database. I blasted two genomes against each other. That is, one query sequence against one database sequence.

If A makes a certain local alignment in B, I would expect to find the same local alignment when I blast B against A. However, as Martin already confirmed, this need not be the case. I still don't know why.

Greetings from Germany,
Michael.
  ----- Original Message ----- 
  From: Pedro Beltrao 
  To: nuhn at rhrk.uni-kl.de 
  Sent: Thursday, January 18, 2007 6:21 PM
  Subject: blast

  I am not sure I understood your question but I would not expect that you would get symmetrical results. Two proteins/genes don't have to necessarily be reciprocal best blast hits. The closest sequence of protein/gene A from species 1 in species 2 could be say protein/gene B but the closest sequence of protein/gene B from species 2 in species 1 could be C. For example if A and C are recently duplicated proteins (in species 1) and C kept the original function while A diverged faster, this will give you the blast results I described. This is why we can use reciprocal best blast hits as a method to predict orthologs. 

  I hope this makes some sense. 
  All the best, 
  Pedro