Reciprocal blast and protein identification

Question

[MAPK]

Hi Guys,

I am doing reciprocal BLAST to identify members of a protein family in different species. Is it necessary for me to set a strict e-value threshold while doing each BLAST? What I am doing is I am setting e-value of 1 in each BLAST search. Once I finish doing reciprocal BLAST, I am planning to do the phylogenetic analysis to further confirm the evolutionarily related gene members in other species. Could you please let me know of anything I am missing in my process of identification?

Thank you.

Answer 1

[Prakki Rama]

Yes. A strict e-value would give more reliable results. Generally, while running BLAST, an e-value of 1e-4 or smaller cutoff is considered. In most of the transcriptome papers, I see people using a cutoff 1e-6 or even much smaller. This is to make much sure that the hit is reliable. On the other hand, if the e-value is big, it is difficult to infer if the hit is really a homolog or not (which somtimes it may be a homolog or sometimes it is not).