Problem With Interpretation Blast Results

Question

[kor_ki]

Hi, I'm struggling with BLAST. I'm not sure if I can properly interpret the results of BLAST. Can anyone tell me when I can be sure that the sequences of the proteins are enough similar to be homologous?

For example: if e-value is 4.3 and identity is 29% - is this homologus or not? What are the criteria? I will be grateful for all the advices. Thank you!

[5heikki]

Significant e-values are in general something less than 1e-6, but really, there's no magic formula that tells you if proteins are homologous or not. Non-homologous proteins more often than not share domains, to which you can obviously get highly significant hits. I suppose you could apply very conservative cut-off criteria, e.g. I would expect that > 90% sequence similarity and > 90% query coverage would return mostly homologs, but would also more than likely miss many true homologs (depending on data).

Answer 1

[kor_ki]

In this case what would be the 'next step' to verify if protein are homologus or not?

[5heikki]

In my opinion, hmmsearch against pfam (or pfam_scan) is a much better way to identify homologs than any blast implementation..